def is_empty(result_frame):
input_size = size
frame_size = result_frame.shape[:2]
image_data = cv2.resize(result_frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
# detect on full image or part of image
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=iou,
score_threshold=score_human
)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = result_frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
summ_of_obj_probabilities = sum([sum(i) for i in bboxes])
return summ_of_obj_probabilities == 0
def run_DL(_frame):
# if pt_cfg.POLYTRACK.DL_DARK_SPOTS:
# dark_spots = pt_cfg.POLYTRACK.RECORDED_DARK_SPOTS
# if len(dark_spots):
# _frame = map_darkspots(_frame, dark_spots)
# else:
# pass
# else:
# pass
_frame = cv2.cvtColor(_frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(_frame)
frame_size = _frame.shape[:2]
image_data = cv2.resize(_frame, (cfg.YOLO.INPUT_SIZE, cfg.YOLO.INPUT_SIZE))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=pt_cfg.POLYTRACK.MAX_OUTPUT_SIZE_PER_CLASS,
max_total_size=pt_cfg.POLYTRACK.MAX_TOTAL_SIZE,
iou_threshold=pt_cfg.POLYTRACK.DL_IOU_THRESHOLD,
score_threshold=pt_cfg.POLYTRACK.DL_SCORE_THRESHOLD)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = _frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
_detections = dl_detections_process(pred_bbox)
return _detections
def get_detected_zone(result_frame, bodyguard=['helmet'], forbidden=False):
input_size = size
frame_size = result_frame.shape[:2]
image_data = cv2.resize(result_frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
# detect on full image or part of image
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=iou,
score_threshold=score_human
)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = result_frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
# if we control emptiness of a room
if forbidden:
return bboxes
obj_detections = []
image = Image.fromarray(result_frame)
for i in range(valid_detections.numpy()[0]):
# save persons parts
image_tmp = image.crop((bboxes[i][0] - 10, bboxes[i][1] - 10, bboxes[i][2] + 10, bboxes[i][3] + 10))
image_tmp = cv2.cvtColor(np.array(image_tmp), cv2.COLOR_BGR2RGB)
obj_detections.append(detect_on_person(image_tmp, bodyguard))
pred_bbox = [bboxes, scores.numpy()[0], classes.numpy()[0], valid_detections.numpy()[0]]
image, violation = utils.draw_bbox(result_frame, pred_bbox, obj_detections, obj_threshold=score_obj)
return image, violation
def detector(images_coming, threshold, prop):
FLAGS(sys.argv)
config = ConfigProto()
config.gpu_options.allow_growth = True
input_size = prop['size']
# load model
saved_model_loaded = tf.saved_model.load(prop['weights'],
tags=[tag_constants.SERVING])
# loop through images in list and run Yolov4 model on each
for count, org_image in enumerate(images_coming, 1):
original_image = cv2.cvtColor(org_image, 1)
image_data = cv2.resize(original_image, (input_size, input_size))
image_data = image_data / 255.
images_data = []
for i in range(1):
images_data.append(image_data)
images_data = np.asarray(images_data).astype(np.float32)
infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
# run non max suppression on detections
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=prop['iou'],
score_threshold=threshold)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
# hold all detection data in one variable
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# if count flag is enabled, perform counting of objects
if prop['count']:
# count objects found
counted_classes = count_objects(pred_bbox,
by_class=True,
allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
image = utils.draw_bbox(original_image,
pred_bbox,
prop['info'],
counted_classes,
allowed_classes=allowed_classes)
else:
image = utils.draw_bbox(original_image,
pred_bbox,
prop['info'],
allowed_classes=allowed_classes)
image = Image.fromarray(image.astype(np.uint8))
image = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2RGB)
return image, counted_classes, pred_bbox
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# get video name by using split method
video_name = video_path.split('/')[-1]
video_name = video_name.split('.')[0]
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
firstFrame = True
frame_num = 0
while True:
return_value, frame_1 = vid.read()
pts = []
aa = []
bb = []
cc = []
dd = []
while firstFrame:
def click_event(event, x, y, flags, param):
global pts
if event == cv2.EVENT_LBUTTONDOWN:
pts.append((x, y))
cv2.circle(frame_1,
center=(x, y),
radius=5,
color=(0, 0, 255),
thickness=-1)
strXY = str(x) + " " + str(y)
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(frame_1, strXY, (x, y), font, 0.5,
(255, 255, 0), 2)
elif event == cv2.EVENT_RBUTTONDOWN:
if pts:
pts.pop()
cv2.imshow('bobur', frame_1)
cv2.imshow('bobur', frame_1)
cv2.setMouseCallback('bobur', click_event)
if cv2.waitKey(1) & 0xFF == ord('c'):
firstFrame = False
break
if len(pts) >= 4:
aa.append(pts[0])
bb.append(pts[1])
cc.append(pts[2])
dd.append(pts[3])
print(aa, bb, cc, dd)
a, b, c, d, e, f, g, h = [209, 1040], [331, 197], [1124, 197], [
1907, 850
], [0, 0], [1920, 0], [1920, 1080], [0, 1080]
# e,f,g,h = [0,0],[1920,0],[1920,1080],[0,1080]
external_poly = [
np.array([e, b, c, f]),
np.array([f, c, d, g]),
np.array([g, d, a, h]),
np.array([h, a, b, e])
]
frame = cv2.fillPoly(frame_1, external_poly, (0, 0, 0))
# cv2.line(frame,(209, 1040),(331,197),(255,0,0),2)
# cv2.line(frame,(331, 197), (1124,197),(255,0,0),2)
# cv2.line(frame,(1124,197),(1907,850),(255,0,0),2)
# cv2.line(frame,(209, 1040),(1907,850),(255,0,0),2)
# cv2.line(frame,a,b,(255,0,0),2)
# cv2.line(frame,b,c,(255,0,0),2)
# cv2.line(frame,c,d,(255,0,0),2)
# cv2.line(frame,a,d,(255,0,0),2)
if return_value:
# frame = cv2.rotate(frame, cv2.ROTATE_90_CLOCKWISE) #rotate the video for mobile videos
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
frame_num += 1
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
if frame_num % 15 == 0:
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(
pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(
pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=100,
max_total_size=100,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h,
original_w)
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# print(pred_bbox[2])
out_boxes, out_scores, out_classes, num_boxes = pred_bbox
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
# allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to allow detections for only SELECTED DETECTION CLASSES)
allowed_classes = ['person', 'car', 'truck', 'bus', 'motorbike']
# allowed_classes = ['car']
# if crop flag is enabled, crop each detection and save it as new image
if FLAGS.crop:
crop_rate = 150 # capture images every so many frames (ex. crop photos every 150 frames)
crop_path = os.path.join(os.getcwd(), 'detections', 'crop',
video_name)
try:
os.mkdir(crop_path)
except FileExistsError:
pass
if frame_num % crop_rate == 0:
final_path = os.path.join(crop_path,
'frame_' + str(frame_num))
try:
os.mkdir(final_path)
except FileExistsError:
pass
crop_objects(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB),
pred_bbox, final_path, allowed_classes)
else:
pass
if FLAGS.count:
# count objects found
counted_classes = count_objects(
pred_bbox, by_class=True, allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
image = utils.draw_bbox(frame,
pred_bbox,
FLAGS.info,
counted_classes,
allowed_classes=allowed_classes,
read_plate=FLAGS.plate)
else:
image = utils.draw_bbox(frame,
pred_bbox,
FLAGS.info,
allowed_classes=allowed_classes,
read_plate=FLAGS.plate)
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(image)
cv2.namedWindow("result", cv2.WINDOW_AUTOSIZE)
result = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("result", result)
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
vid.release()
cv2.destroyAllWindows()
max_total_size=50,
iou_threshold=0.45,
score_threshold=0.50)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# get video name by using split method
video_name = video_path.split('/')[-1]
video_name = video_name.split('.')[0]
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights, tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
frame_num = 0
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
frame_num += 1
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [interpreter.get_tensor(output_details[i]['index']) for i in range(len(output_details))]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1], pred[0], score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0], pred[1], score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score
)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
pred_bbox = [bboxes, scores.numpy()[0], classes.numpy()[0], valid_detections.numpy()[0]]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to allow detections for only people)
#allowed_classes = ['person']
# if crop flag is enabled, crop each detection and save it as new image
if FLAGS.crop:
crop_rate = 150 # capture images every so many frames (ex. crop photos every 150 frames)
crop_path = os.path.join(os.getcwd(), 'detections', 'crop', video_name)
try:
os.mkdir(crop_path)
except FileExistsError:
pass
if frame_num % crop_rate == 0:
final_path = os.path.join(crop_path, 'frame_' + str(frame_num))
try:
os.mkdir(final_path)
except FileExistsError:
pass
crop_objects(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB), pred_bbox, final_path, allowed_classes)
else:
pass
if FLAGS.count:
# count objects found
counted_classes = count_objects(pred_bbox, by_class = True, allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
image = utils.draw_bbox(frame, pred_bbox, FLAGS.info, counted_classes, allowed_classes=allowed_classes, read_plate=FLAGS.plate)
else:
image = utils.draw_bbox(frame, pred_bbox, FLAGS.info, allowed_classes=allowed_classes, read_plate=FLAGS.plate)
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(image)
# cv2.namedWindow("result", cv2.WINDOW_AUTOSIZE)
result = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
# if not FLAGS.dont_show:
# cv2.imshow("result", result)
if FLAGS.output:
out.write(result)
def iterate(lines, model, vid, frame_num):
tracks = []
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
cv2.destroyAllWindows()
return False, tracks
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (FLAGS.size, FLAGS.size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
# run detections using yolov3 if flag is set
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[FLAGS.size, FLAGS.size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[FLAGS.size, FLAGS.size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = model.signatures['serving_default'](batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
#allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
#allowed_classes = ['person']
allowed_classes = ['car', 'bus', 'truck']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(frame, "Objects being tracked: {}".format(count), (5, 35),
cv2.FONT_HERSHEY_COMPLEX_SMALL, 2, (0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name,
feature) for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
tracks.append(
Rect(track.track_id, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]) - int(bbox[0]), int(bbox[3]) - int(bbox[1]))))
# draw bbox on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17, int(bbox[1])),
color, -1)
cv2.putText(frame, class_name + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
# if enable info flag then print details about each track
#if FLAGS.info:
# print("Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}".format(str(track.track_id), class_name, (int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
for line in lines:
cv2.line(frame, line.pt1, line.pt2, line.color, 3)
cv2.line(frame, line.vertor_pt1, line.vertor_pt2, (255, 255, 0), 2)
cv2.putText(frame, str(line.count), line.center,
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
cv2.putText(frame, str(frame_num), (50, 50), cv2.FONT_HERSHEY_SIMPLEX, 1,
(255, 255, 0), 2)
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'):
cv2.destroyAllWindows()
return False, tracks
return True, tracks
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
allowed_classes = ['person']
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
image = utils.draw_bbox(frame,
pred_bbox,
allowed_classes=allowed_classes)
if FLAGS.covid:
distance = social_distance(pred_bbox, frame, allowed_classes)
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(image)
cv2.namedWindow("result", cv2.WINDOW_AUTOSIZE)
result = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("result", result)
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
cv2.destroyAllWindows()
def detect_video(url):
"""
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = 416
saved_model_loaded = tf.saved_model.load('./checkpoints/yolov4Tiny-416', tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
"""
try:
name = random.random()
vid = cv2.VideoCapture(url)
# out = None
currentFrame = 0
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
currentFrame += 1
if ((currentFrame % 4) == 0) & ((currentFrame % 15) != 0):
continue
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=0.45,
score_threshold=0.25
)
#bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
pred_bbox = [boxes.numpy(), scores.numpy(), classes.numpy(), valid_detections.numpy()]
if (currentFrame % 15) == 0:
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
predictions = [bboxes, scores.numpy()[0], classes.numpy()[0], valid_detections.numpy()[0]]
crop_detections(frame, predictions, 200, 50)
image = utils.draw_bbox(frame, pred_bbox)
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(image)
cv2.namedWindow(str(id), cv2.WINDOW_AUTOSIZE)
result = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
cv2.imshow(str(name), result)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cv2.destroyAllWindows()
return None
except Exception as e:
print('oof', e)
cv2.destroyAllWindows()
return None
def inference(preprocess_queue, inference_queue):
import tensorflow as tf
import core.utils as utils
from tensorflow.python.saved_model import tag_constants
from tensorflow.compat.v1 import InteractiveSession
from tensorflow.compat.v1 import ConfigProto
from core.functions import count_objects, crop_objects
from core.config import cfg
from core.utils import read_class_names
import os
import random
from core.yolov4 import filter_boxes
tf.keras.backend.clear_session()
input_size = Parameters.input_size
model = OutsourceContract.model
framework = Parameters.framework
tiny = OutsourceContract.tiny
weights = Parameters.weights
iou = Parameters.iou
score = Parameters.score
physical_devices = tf.config.experimental.list_physical_devices('GPU')
try:
if len(physical_devices) > 0:
tf.config.experimental.set_memory_growth(physical_devices[0], True)
except:
pass
# configure gpu usage
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
# load model
if framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=weights)
else:
saved_model_loaded = tf.saved_model.load(weights,
tags=[tag_constants.SERVING])
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
count = Parameters.count
info = Parameters.info
crop = Parameters.crop
while True:
if not preprocess_queue.empty():
queueData = preprocess_queue.get()
while not preprocess_queue.empty():
queueData = preprocess_queue.get()
#preprocess_queue.task_done()
images_data = queueData[0]
name = queueData[1]
original_image = queueData[2]
#preprocess_queue.task_done()
if framework == 'tflite':
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
interpreter.set_tensor(input_details[0]['index'], images_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
if model == 'yolov3' and tiny == True:
boxes, pred_conf = filter_boxes(
pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(
pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=iou,
score_threshold=score) # 1.2ms
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h,
original_w) # 1ms #-> no tf needed
# hold all detection data in one variable
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to allow detections for only people)
# allowed_classes = ['person']
# if crop flag is enabled, crop each detection and save it as new image
if crop:
crop_path = os.path.join(os.getcwd(), 'detections', 'crop',
image_name)
try:
os.mkdir(crop_path)
except FileExistsError:
pass
crop_objects(cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB),
pred_bbox, crop_path, allowed_classes)
if count:
# count objects found
counted_classes = count_objects(
pred_bbox, by_class=False, allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
boxtext, image = utils.draw_bbox(
original_image,
pred_bbox,
info,
counted_classes,
allowed_classes=allowed_classes)
else:
boxtext, image = utils.draw_bbox(
original_image,
pred_bbox,
info,
allowed_classes=allowed_classes) # 0.5ms
image = Image.fromarray(image.astype(np.uint8)) # 0.3ms
inference_queue.put((boxtext, image, name))
def main(_argv):
avg=[]
# Definition of the parameters
max_cosine_distance = 0.4
nn_budget = None
nms_max_overlap = 1.0
#regression model load
weight_path='./2_input_model_2-3.5%/'
loaded_model = tf.keras.models.load_model(weight_path)
# initialize deep sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# load tflite model if flag is set
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights, tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
# get video ready to save locally if flag is set
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
frame_num = 0
# while video is running
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_num +=1
print('Frame #: ', frame_num)
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [interpreter.get_tensor(output_details[i]['index']) for i in range(len(output_details))]
# run detections using yolov3 if flag is set
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1], pred[0], score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0], pred[1], score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score
)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
#print("pred_bbox: ",pred_bbox[0])
#print("scores: ",pred_bbox[1])
#print("classes :",pred_bbox[2])
#print("num :",pred_bbox[3])
#print("width :",width)
#print("height :",height)
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
#allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(frame, "Objects being tracked: {}".format(count), (5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2, (0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [Detection(bbox, score, class_name, feature) for bbox, score, class_name, feature in zip(bboxes, scores, names, features)]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
#print("boxs ",boxs)
#print("scores ",scores)
#print("classes ",classes)
indices = preprocessing.non_max_suppression(boxs, classes, nms_max_overlap, scores)
#print("indices ",indices)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
cv2.putText(frame, "using regress", (5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2, (255, 0, 255), 2)
#cv2.putText(frame, "Objects being detected: {}".format(count), (5, 350), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2, (0, 0, 255), 2)
cv2.putText(frame, "frame# {}".format(frame_num), (750, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2, (255, 0, 255), 2)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
if 'entrance' not in classes:
if len(classes)>1:
if(contains_duplicates(classes)==False):
#color = (50, 89, 170)
check_rect=0
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
########## set sticker as low priority#############
if ((classes[0]=='mat' or 'sensor') and (classes[1]=='mat' or 'sensor')):
print("*************NO STK**********************************")
color = (50, 89, 170)
x1,y1,x2,y2=convert2(width,height,int(boxs[0][0]),int(boxs[0][1]),int(boxs[0][0]+boxs[0][2]),int(boxs[0][1]+boxs[0][3]))#xywh to xmin ymin xmax ymax
x3,y3,x4,y4=convert2(width,height,int(bboxes[1][0]),int(bboxes[1][1]),int(bboxes[1][0]+bboxes[1][2]),int(bboxes[1][1]+bboxes[1][3]))#xywh to xmin ymin xmax ymax
reg_input=np.array([[class_index(classes[0]),x1,y1,x2,y2,class_index(classes[1]),x3,y3,x4,y4]])
predictions = loaded_model.predict(reg_input)
a1_pred = predictions[0]
b1_pred = predictions[1]
c1_pred = predictions[2]
d1_pred = predictions[3]
xmin,xmax,ymin,ymax=convert(width,height,a1_pred,b1_pred,c1_pred,d1_pred)
start_point = (xmin, ymin)
end_point = (xmax, ymax)
rect1=xmax-xmin
rect2=ymax-ymin
check_rect=rect2/rect1
################ else condition for sticker ######
else:
print("*************USE STK**********************************")
if ((classes[0]=='famSticker' or 'okmartSticker' or 'sevenSticker')):
color = (60, 120, 40)
x1,y1,x2,y2=convert2(width,height,int(boxs[0][0]),int(boxs[0][1]),int(boxs[0][0]+boxs[0][2]),int(boxs[0][1]+boxs[0][3]))#xywh to xmin ymin xmax ymax
x3,y3,x4,y4=convert2(width,height,int(bboxes[1][0]),int(bboxes[1][1]),int(bboxes[1][0]+bboxes[1][2]),int(bboxes[1][1]+bboxes[1][3]))#xywh to xmin ymin xmax ymax
reg_input=np.array([[class_index(classes[0]),x1,y1,x2,y2,class_index(classes[1]),x3,y3,x4,y4]])
#### rario ####
C1_x=boxs[0][0]+(boxs[0][2]/2)
C1_y=boxs[0][1]+(boxs[0][3]/2)
C2_x=bboxes[1][0]+(bboxes[1][2]/2)
C2_y=bboxes[1][1]+(bboxes[1][3]]/2)
Dx = (C2_x - C1_x);
Dy = (C2_y - C1_y);
#The two rectangles do not intersect, and there are two rectangles partially overlapping in the X-axis direction. The minimum distance is the distance between the lower line of the upper rectangle and the upper line of the lower rectangle
if((Dx < ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)) && (Dy >= ((int(boxs[0][1]+boxs[0][3]) + rint(bboxes[1][1]+bboxes[1][3])) / 2))):
min_dist = Dy - ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3])) / 2);
#The two rectangles do not intersect. There are two partially overlapping rectangles in the Y-axis direction. The minimum distance is the distance between the right line of the left rectangle and the left line of the right rectangle
elif((Dx >= ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)) && (Dy < ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3])) / 2))):
min_dist = Dx - ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2);
#Two rectangles do not intersect, two rectangles that do not overlap in the X-axis and Y-axis directions, the minimum distance is the distance between the two closest vertices,
# Using the Pythagorean theorem, it is easy to calculate this distance
elif((Dx >= ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)) && (Dy >= ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3])) / 2))):
int delta_x = Dx - ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2);
int delta_y = Dy - ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3]))/ 2);
min_dist = sqrt(delta_x * delta_x + delta_y * delta_y);
#The intersection of two rectangles, the minimum distance is negative, return -1
else:
min_dist = -1;
if(classes[1]=='mat'):
if((min_dist/Dy)<3):
predictions = loaded_model.predict(reg_input)
a1_pred = predictions[0]
b1_pred = predictions[1]
c1_pred = predictions[2]
d1_pred = predictions[3]
xmin,xmax,ymin,ymax=convert(width,height,a1_pred,b1_pred,c1_pred,d1_pred)
start_point = (xmin, ymin)
end_point = (xmax, ymax)
rect1=xmax-xmin
rect2=ymax-ymin
check_rect=rect2/rect1
else:
print("not predict")
elif(classes[1]=='sensor'):
if((min_dist/Dx)<3):
predictions = loaded_model.predict(reg_input)
a1_pred = predictions[0]
b1_pred = predictions[1]
c1_pred = predictions[2]
d1_pred = predictions[3]
xmin,xmax,ymin,ymax=convert(width,height,a1_pred,b1_pred,c1_pred,d1_pred)
start_point = (xmin, ymin)
end_point = (xmax, ymax)
rect1=xmax-xmin
rect2=ymax-ymin
check_rect=rect2/rect1
else:
print("not predict")
elif((classes[1]=='famSticker' or 'okmartSticker' or 'sevenSticker')):
color = (60, 120, 40)
x1,y1,x2,y2=convert2(width,height,int(boxs[0][0]),int(boxs[0][1]),int(boxs[0][0]+boxs[0][2]),int(boxs[0][1]+boxs[0][3]))#xywh to xmin ymin xmax ymax
x3,y3,x4,y4=convert2(width,height,int(bboxes[1][0]),int(bboxes[1][1]),int(bboxes[1][0]+bboxes[1][2]),int(bboxes[1][1]+bboxes[1][3]))#xywh to xmin ymin xmax ymax
reg_input=np.array([[class_index(classes[0]),x1,y1,x2,y2,class_index(classes[1]),x3,y3,x4,y4]])
#### rario ####
C1_x=boxs[0][0]+(boxs[0][2]/2)
C1_y=boxs[0][1]+(boxs[0][3]/2)
C2_x=bboxes[1][0]+(bboxes[1][2]/2)
C2_y=bboxes[1][1]+(bboxes[1][3]/2)
Dx = (C2_x - C1_x)
Dy = (C2_y - C1_y)
#The two rectangles do not intersect, and there are two rectangles partially overlapping in the X-axis direction. The minimum distance is the distance between the lower line of the upper rectangle and the upper line of the lower rectangle
if((Dx < ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)) && (Dy >= ((int(boxs[0][1]+boxs[0][3]) + rint(bboxes[1][1]+bboxes[1][3])) / 2))):
min_dist = Dy - ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3])) / 2)
#The two rectangles do not intersect. There are two partially overlapping rectangles in the Y-axis direction. The minimum distance is the distance between the right line of the left rectangle and the left line of the right rectangle
elif((Dx >= ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)) && (Dy < ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3])) / 2))):
min_dist = Dx - ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)
#Two rectangles do not intersect, two rectangles that do not overlap in the X-axis and Y-axis directions, the minimum distance is the distance between the two closest vertices,
# Using the Pythagorean theorem, it is easy to calculate this distance
elif((Dx >= ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)) && (Dy >= ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3])) / 2))):
int delta_x = Dx - ((int(boxs[0][0]+boxs[0][2]) + int(bboxes[1][0]+bboxes[1][2]))/ 2)
int delta_y = Dy - ((int(boxs[0][1]+boxs[0][3]) + int(bboxes[1][1]+bboxes[1][3]))/ 2)
min_dist = sqrt(delta_x * delta_x + delta_y * delta_y)
#The intersection of two rectangles, the minimum distance is negative, return -1
else:
min_dist = -1
if(classes[0]=='mat'):
if((min_dist/Dy)<3):
predictions = loaded_model.predict(reg_input)
a1_pred = predictions[0]
b1_pred = predictions[1]
c1_pred = predictions[2]
d1_pred = predictions[3]
xmin,xmax,ymin,ymax=convert(width,height,a1_pred,b1_pred,c1_pred,d1_pred)
start_point = (xmin, ymin)
end_point = (xmax, ymax)
rect1=xmax-xmin
rect2=ymax-ymin
check_rect=rect2/rect1
else:
print("not predict")
elif(classes[0]=='sensor'):
if((min_dist/Dx)<3):
predictions = loaded_model.predict(reg_input)
a1_pred = predictions[0]
b1_pred = predictions[1]
c1_pred = predictions[2]
d1_pred = predictions[3]
xmin,xmax,ymin,ymax=convert(width,height,a1_pred,b1_pred,c1_pred,d1_pred)
start_point = (xmin, ymin)
end_point = (xmax, ymax)
rect1=xmax-xmin
rect2=ymax-ymin
check_rect=rect2/rect1
else:
print("not predict")
##############
##########################################
######## check door size and display #########if check_rect>1 and frame_num !=104:
print("check_rect:{}".format(check_rect))
if check_rect>1 :
blk = np.zeros(frame.shape, np.uint8)
cv2.rectangle(blk, start_point, end_point, color, cv2.FILLED)
frame =cv2.addWeighted(frame, 1.0, blk, 0.5, 1)
print("predict_BBox Coords (xmin, ymin, xmax, ymax): {}".format((xmin,ymin,xmax,ymax)))
else:
print("not show predicted bbox")
###############################
########
# select one entrace
########
#if classes.count('entrance')>1:
# entrance_num=[]
# iou_list=[]
# iou_check=[]
# for i in range(len(classes)):
# if classes[i]=='entrance'
# entrance_num.append(i)
# if len(classes)>1:
# if(contains_duplicates(classes)==False):
# color = (50, 89, 170)
# width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
# height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
# x1,y1,x2,y2=convert2(width,height,int(boxs[0][0]),int(boxs[0][1]),int(boxs[0][0]+boxs[0][2]),int(boxs[0][1]+boxs[0][3]))#xywh to xmin ymin xmax ymax
# x3,y3,x4,y4=convert2(width,height,int(bboxes[1][0]),int(bboxes[1][1]),int(bboxes[1][0]+bboxes[1][2]),int(bboxes[1][1]+bboxes[1][3]))#xywh to xmin ymin xmax ymax
# reg_input=np.array([[class_index(classes[0]),x1,y1,x2,y2,class_index(classes[1]),x3,y3,x4,y4]])
# predictions = loaded_model.predict(reg_input)
# a1_pred = predictions[0]
# b1_pred = predictions[1]
# c1_pred = predictions[2]
# d1_pred = predictions[3]
# xmin,xmax,ymin,ymax=convert(width,height,a1_pred,b1_pred,c1_pred,d1_pred)
# ###IOU###
# GT_bbox_area = (xmax - xmin + 1) * ( ymax -ymin + 1)
# ###########
# ##check entrace##
# Pred_bbox_area =(x_bottomright_p - x_topleft_p + 1 ) * ( y_bottomright_p -y_topleft_p + 1)
# x_top_left =np.max([x_topleft_gt, x_topleft_p])
# y_top_left = np.max([y_topleft_gt, y_topleft_p])
# x_bottom_right = np.min([x_bottomright_gt, x_bottomright_p])
# y_bottom_right = np.min([y_bottomright_gt, y_bottomright_p])
#
# intersection_area = (x_bottom_right- x_top_left + 1) * (y_bottom_right-y_top_left + 1)
#
# union_area = (GT_bbox_area + Pred_bbox_area - intersection_area)
#
# iou_check.append(intersection_area/union_area)
#
# for j in len(iou_check):
# if entrance_num[j]<iou_check.max:
# track.delete
#if(int(track.track_id)>=3 or (int(track.track_id)>10 and int(track.track_id)<20 ) ):
#frame_num
###################### draw bbox on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
if(class_name=='entrance'):
if( int(track.track_id)==1 and frame_num>121):
print("skip Tracker ID: {}, Class: {}".format(str(track.track_id), class_name))
else:
print("RED Tracker ID: {}, Class: {}".format(str(track.track_id), class_name))
blk = np.zeros(frame.shape, np.uint8)
cv2.rectangle(blk,(int(bbox[0]*1.05), int(bbox[1]*1.05)), (int(bbox[2]*0.95), int(bbox[3]*0.95)), (255, 0, 0), cv2.FILLED)
frame =cv2.addWeighted(frame, 1.0, blk, 0.5, 1)
cv2.rectangle(frame, (int(bbox[0]*1.05), int(bbox[1]*1.05)), (int(bbox[2]*0.95), int(bbox[3]*0.95)), color, 2)
cv2.rectangle(frame, (int(bbox[0]*1.05), int(bbox[1]*1.05-30)), (int(bbox[0]*1.05)+(len(class_name)+len(str(track.track_id)))*17, int(bbox[1]*1.05)), color, -1)
cv2.putText(frame, class_name + "-" + str(track.track_id),(int(bbox[0]*1.05), int(bbox[1]*1.05-10)),0, 0.75, (255,255,255),2)
# if enable info flag then print details about each track
if FLAGS.info:
print("Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}".format(str(track.track_id), class_name, (int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
avg.append(fps)
print("avg fps {}".format(statistics.mean(avg)))
cv2.putText(frame, "FPS: %.2f" % fps, (50, 500), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2, (66, 245, 141), 2)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
def detect(saved_model_loaded, infer, input_size, image_path):
# config = ConfigProto()
# session = InteractiveSession(config=config)
# STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
# start = time.perf_counter()
original_image = cv2.imread(image_path)
original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
image_data = cv2.resize(original_image, (input_size, input_size))
image_data = image_data / 255.
count = 0
# get image name by using split method
image_name = image_path.split('/')[-1]
image_name = image_name.split('.')[0]
images_data = []
for i in range(1):
images_data.append(image_data)
images_data = np.asarray(images_data).astype(np.float32)
#infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
# run non max suppression on detections
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=0.20,
score_threshold=0.20)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
# hold all detection data in one variable
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
ocr(cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB), pred_bbox)
# count objects found
counted_classes = count_objects(pred_bbox,
by_class=True,
allowed_classes=['person'])
# print(counted_classes.items())
if len(counted_classes.items()) != 0:
count += counted_classes['person']
else:
count += 0
# image = utils.draw_bbox(original_image, pred_bbox, False, counted_classes, allowed_classes=['person'])
# image = Image.fromarray(image.astype(np.uint8))
# image.show()
# image = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2RGB)
# cv2.imwrite(FLAGS.output + 'detection' + str(count) + '.png', image)
update_db(count)
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
image_path = FLAGS.image
original_image = cv2.imread(image_path)
original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
# image_data = utils.image_preprocess(np.copy(original_image), [input_size, input_size])
image_data = cv2.resize(original_image, (input_size, input_size))
image_data = image_data / 255.
# image_data = image_data[np.newaxis, ...].astype(np.float32)
images_data = []
for i in range(1):
images_data.append(image_data)
images_data = np.asarray(images_data).astype(np.float32)
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
interpreter.set_tensor(input_details[0]['index'], images_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
# hold all detection data in one variable
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# pred_bbox = [boxes.numpy(), scores.numpy(), classes.numpy(), valid_detections.numpy()]
# image = utils.draw_bbox(original_image, pred_bbox)
# image = utils.draw_bbox(image_data*255, pred_bbox)
if FLAGS.count:
# count objects found
counted_classes = count_objects(pred_bbox,
by_class=True,
allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
image = utils.draw_bbox(original_image,
pred_bbox,
FLAGS.info,
counted_classes,
allowed_classes=allowed_classes)
else:
image = utils.draw_bbox(original_image,
pred_bbox,
FLAGS.info,
allowed_classes=allowed_classes)
image = Image.fromarray(image.astype(np.uint8))
image.show()
image = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2RGB)
cv2.imwrite(FLAGS.output, image)
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# get video name by using split method
video_name = video_path.split('/')[-1]
video_name = video_name.split('.')[0]
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights, tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
frame_num = 0
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
frame_num += 1
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [interpreter.get_tensor(output_details[i]['index']) for i in range(len(output_details))]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1], pred[0], score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0], pred[1], score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score
)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
pred_bbox = [bboxes, scores.numpy()[0], classes.numpy()[0], valid_detections.numpy()[0]]
# print(pred_bbox[2])
out_boxes, out_scores, out_classes, num_boxes = pred_bbox
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
# allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to allow detections for only SELECTED DETECTION CLASSES)
allowed_classes = ['car','truck','motorbike','bus']
# allowed_classes = ['car']
#################################################################################################################################
#Calculting the distance xmin, ymin, xmax, ymax
cv2.line(img=frame, pt1=(595,940),pt2=(1567,940), color=(0, 0, 180), thickness=3, lineType=8, shift=0)
for i, b in zip(out_boxes,out_classes):
f = 1460 ## Focal length of the camera
if i[0]>1 and i[1]>105 and i[2]<1920 and i[3]<910 and int(b) == 2:
wpix = i[2] - i[0]
w = 1.7 #car width
# D = round((f*w)/wpix, 2)
# h = 1.2 #ORIGINAL === 1.6 # Most vehicles have a size that ranges from 1.5 – 1.8 meters high and widths of 1.6-1.7 meters.
d_original =round((f*w)/wpix,2)
d = d_original - 3 #2.3 from the camera to the front of the car 2.2 is for yolo car
d = round(d,2)
print("{} meters".format(d_original),end = ",")
print("{} meters".format(d))
cv2.putText(frame, "{}m".format(d), (int(i[0]+ ((int(i[2]-int(i[0]))/2))), int(i[1])-20), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (36,255,12), 2)
elif i[0]>1 and i[1]>105 and i[2]<1920 and i[3]<940 and int(b) == 3:
# wpix = i[3] - i[1]
# w = 1.6 #car width
# D = round((f*w)/wpix, 2)
h = 0.75 # Most motorbikes height 75 cm
d_original =round((f*h)/wpix,2)
d = d_original - 2.3 -1.1 # 2.3 from the camera to the front of the car 1.1 is for yolo motorbike
d = round(d,2)
print("{} meters".format(d_original),end = ",")
print("{} meters".format(d))
cv2.putText(frame, "{}m".format(d), (int(i[0]+ ((int(i[2]-int(i[0]))/2))), int(i[1])-20), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (36,255,12), 2)
# ##Because bus is too long I will measure with width
elif i[0]>1 and i[1]>105 and i[2]<1920 and i[3]<940 and int(b) == 5:
wpix = i[2] - i[0]
# w = 1.6 #
# D = round((f*w)/wpix, 2)
# h = 3.5 # Most buses height 4.3 meters
w = 2.3
d_original =round((f*w)/wpix,2)
d = d_original - 2.3 # 2.3 from the camera to the front of the car BUSSS
d = round(d,2)
print("{} meters".format(d_original),end = ",")
print("{} meters".format(d))
cv2.putText(frame, "{}m".format(d), (int(i[0]+ ((int(i[2]-int(i[0]))/2))), int(i[1])-20), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (36,255,12), 2)
elif i[0]>1 and i[1]>105 and i[2]<1920 and i[3]<940 and int(b) == 6:
wpix = i[3] - i[1]
# w = 1.6 #car width
# D = round((f*w)/wpix, 2)
h = 4.3 # Most train height 4.3 meters
d_original =round((f*h)/wpix,2)
d = d_original - 2.3 # 2.3 from the camera to the front of the car, NO TRAINNNNNNNNN
d = round(d,2)
print("{} meters".format(d_original),end = ",")
print("{} meters".format(d))
cv2.putText(frame, "{}m".format(d), (int(i[0]+ ((int(i[2]-int(i[0]))/2))), int(i[1])-20), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (36,255,12), 2)
# #Because truck is too long I will measure with width
elif i[0]>1 and i[1]>105 and i[2]<1920 and i[3]<910 and int(b) == 7:
wpix = i[3] - i[1]
# w = 2.7 #Most truck height 2.7
# D = round((f*w)/wpix, 2)
h = 3 # Most truck height 4.3 meters
# w = 1.8
d_original =round((f*h)/wpix,2)
d = d_original - 3 # 2.3 from the camera to the front of the car 5 is for yolo TRUCKCKKKKKKK
d = round(d,2)
print("{} meters".format(d_original),end = ",")
print("{} meters".format(d))
cv2.putText(frame, "{}m".format(d), (int(i[0]+ ((int(i[2]-int(i[0]))/2))), int(i[1])-20), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (36,255,12), 2)
else:
None
###################################################################################################################################
# if crop flag is enabled, crop each detection and save it as new image
if FLAGS.crop:
crop_rate = 150 # capture images every so many frames (ex. crop photos every 150 frames)
crop_path = os.path.join(os.getcwd(), 'detections', 'crop', video_name)
try:
os.mkdir(crop_path)
except FileExistsError:
pass
if frame_num % crop_rate == 0:
final_path = os.path.join(crop_path, 'frame_' + str(frame_num))
try:
os.mkdir(final_path)
except FileExistsError:
pass
crop_objects(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB), pred_bbox, final_path, allowed_classes)
else:
pass
if FLAGS.count:
# count objects found
counted_classes = count_objects(pred_bbox, by_class = False, allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
image = utils.draw_bbox(frame, pred_bbox, FLAGS.info, counted_classes, allowed_classes=allowed_classes, read_plate=FLAGS.plate)
else:
image = utils.draw_bbox(frame, pred_bbox, FLAGS.info, allowed_classes=allowed_classes, read_plate=FLAGS.plate)
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(image)
cv2.namedWindow("result", cv2.WINDOW_AUTOSIZE)
result = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("result", result)
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
cv2.destroyAllWindows()
def main(_argv):
# Definition of the parameters
max_cosine_distance = 0.5
nn_budget = None
nms_max_overlap = 0.8
# initialize deep sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
# Loading the pretrained model weights
saved_model_loaded = tf.saved_model.load(FLAGS.weights_path,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# Loading the stores configuration JSON file
stores_config_filename = 'stores_sections.json'
stores_config_filepath = os.path.join(
os.path.dirname(os.path.abspath(__file__)), stores_config_filename)
stores_sections = load_json_file(stores_config_filepath)
#Creating engine to query from the database
engine = get_db()
# Getting the list of video filenames that had been processed
processed_videos = pd.read_sql('SELECT DISTINCT name_video FROM counts',
engine)
processed_videos = processed_videos.name_video.tolist()
# Get the current directory where this file person_tracker.py is located
file_directory = os.getcwd()
# Changing to the root directory (Google Colab root directory in this case)
# to be able to extract the videofilenames in another location
os.chdir('/content')
# Getting the video filenames available on the repository
mypath = FLAGS.videos_repository_path
onlyfiles = [
f for f in os.listdir(mypath)
if os.path.isfile(os.path.join(mypath, f))
]
# Changing back to the person_tracker.py directory to continue with the process of the videos
os.chdir(file_directory)
# Computing the video filanames that need to be processed
videos_to_process = list(set(onlyfiles) - set(processed_videos))
print("Videos to process: ", len(videos_to_process))
# Loop for process all the videos that had not been processed
for i in range(0, len(videos_to_process)):
print(f"Processing video: {i+1}/{len(videos_to_process)}")
# Initializing variables from the Flags values
input_size = FLAGS.size
video_path = os.path.join(mypath, videos_to_process[i])
print(video_path)
output_csv_path = FLAGS.output_csv_path
count_csv_path = FLAGS.count_csv_path
file_name = video_path.split("/")[-1]
#Extract the date and time information and camera number from the video_path string
if len(re.findall('[0-9]{14}', video_path)) == 2:
time_start_vid, time_end_vid = re.findall('[0-9]{14}', video_path)
time_start_vid_dt = datetime.strptime(str(time_start_vid),
'%Y%m%d%H%M%S')
time_end_vid_dt = datetime.strptime(str(time_end_vid),
'%Y%m%d%H%M%S')
camera = int(re.findall(r'_([0-9]{1})_', video_path.lower())[0])
# Limit line points for the people counter
# the only cameras of interest to count people in and out is camera 1 and camera 2
if camera == 1:
startline = (614, 95)
endline = (807, 95)
elif camera == 2:
startline = (305, 175)
endline = (476, 175)
else:
startline = (0, 0)
endline = (0, 0)
# Extract the name of the store from the video_path string
store_name = re.findall(r'/([a-z0-9\s]*)_*', video_path.lower())[-1]
# Change the default name that video filenames have of san diego store
if store_name == 'hermeco oficinas':
store_name = 'san diego'
# Begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
# Get video features
out = None
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
frame_count = int(vid.get(
cv2.CAP_PROP_FRAME_COUNT)) # Total number of frames in the video
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
delta_time = (time_end_vid_dt - time_start_vid_dt) / frame_count
# get video ready to save locally if flag is set
if FLAGS.output_vid:
out = cv2.VideoWriter(FLAGS.output_vid, codec, fps,
(width, height))
frame_num = 1
# Initialize the fields of the dataframe that will store the detections
detections_df = pd.DataFrame({
'Store_name': [],
'Start_date': [],
'End_date': [],
'current_datetime': [],
'Camera': [],
'Object': [],
'Id': [],
'X_center_original': [],
'Y_center_original': [],
'X_center_perspective': [],
'Y_center_perspective': [],
'X_min': [],
'Y_min': [],
'X_max': [],
'Y_max': [],
'Frame': []
})
temp = pd.DataFrame()
# vector that will store the las 15 locations of each track
pts = [deque(maxlen=15) for _ in range(10000)]
counter_out = []
counter_in = []
start_process = time.time()
# while video is running
while True:
# Get the frame image from the video
return_value, frame = vid.read()
# transform the defailt color of OpenCV frame BGR to RGB
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print(
'Video has ended or failed, try a different video format!')
break
#print('Frame #: ', frame_num)
# Preprocessing the frame image
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
# Getting all the bounding boxes of the detections and their respective confidence
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
# Applying Non-maximum Suppression to get the best bounding box for each detection
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score_th)
# Convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
#allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.display_count:
cv2.putText(frame, "Objects being tracked: {}".format(count),
(5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2,
(0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(
boxs, classes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# Computing the time has passed since the beggining of the video to the current frame
delta_time_frame = delta_time * (frame_num - 1)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
# draw bbox on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(frame, class_name + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
# if enable info flag then print details about each track
if FLAGS.info:
print(
f"Tracker ID: {str(track.track_id)}, Class: {class_name}, BBox Coords (xmin, ymin, xmax, ymax): {(int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))}"
)
# computing the bottom center of the bounding box
center = (int(((bbox[0]) + (bbox[2])) / 2), int(bbox[3]))
# Loop for the diferent configured sections of the store for the current camera
for sec in stores_sections[store_name][
f'camera_{camera}'].keys():
# Get the 4 points of the section
bound_section_points = stores_sections[store_name][
f'camera_{camera}'][sec]['camera_view_points']
# Verify if the center point of the detection is in the section region
mpltPath_path = mpltPath.Path(bound_section_points)
inside = mpltPath_path.contains_point(list(center))
if inside:
# Get the perspective transformation matrix
transform_matrix = np.array(
stores_sections[store_name][f'camera_{camera}']
[sec]['transformation_matrix'])
# Apply the transformation matrix to transform the point to the blueprint perspective
transformed_center = point_perspective_transform(
center, transform_matrix)[0]
break
else:
transformed_center = [0, 0]
# Appending the center to the corrent track
pts[track.track_id].append(center)
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), color, thickness)
height, width, _ = frame.shape
##cv2.line(frame,(0,int(3*height/6)),(width,int(3*height/6)),(0,0,255), thickness = 2)
#cv2.line(frame,(193,183),(650,183),(0,0,255),2)
cv2.line(frame, startline, endline, (0, 0, 255), 2)
# split the bounding box bottom center coordinates
center_x = center[0]
if (camera == 2):
center_y = int(bbox[3])
else:
center_y = int(((bbox[1]) + (bbox[3])) / 2)
# Counting if the track is leaving or entering the camera section
# based in the direction the person is crossing a fixed line
if (center_y <= int(startline[1] + 20)) and (
center_y >= int(startline[1] - 20)) and (
center_x >= int(startline[0] - 30)) and (
center_x <= int(endline[0] + 30)):
if class_name == 'person':
list_y = [i[1] for i in pts[track.track_id]]
in_var = all(x < y for x, y in zip(list_y, list_y[1:]))
out_var = all(x > y
for x, y in zip(list_y, list_y[1:]))
if in_var and len(list_y) > 1:
counter_in.append(int(track.track_id))
elif out_var and len(list_y) > 1:
counter_out.append(int(track.track_id))
# Adding the current track detection data to the dataframe
temp = pd.DataFrame({
'Store_name': [store_name],
'Start_date': [time_start_vid_dt],
'End_date': [time_end_vid_dt],
'current_datetime': [time_start_vid_dt + delta_time_frame],
'Camera': [int(camera)],
'Object': [class_name],
'Id': [int(track.track_id)],
'X_center_original': [int(center[0])],
'Y_center_original': [int(center[1])],
'X_center_perspective': [int(transformed_center[0])],
'Y_center_perspective': [int(transformed_center[1])],
'X_min': [int(bbox[0])],
'Y_min': [int(bbox[1])],
'X_max': [int(bbox[2])],
'Y_max': [int(bbox[3])],
'Frame': [int(frame_num)]
})
detections_df = pd.concat([detections_df, temp],
ignore_index=True)
# Getting the total in and out counts
total_count_in = len(set(counter_in))
total_count_out = len(set(counter_out))
cv2.putText(frame, 'Total Count In:' + str(len(set(counter_in))),
(0, 130), 0, 1, (0, 0, 255), 2)
cv2.putText(frame, 'Total Count Out:' + str(len(set(counter_out))),
(0, 200), 0, 1, (0, 0, 255), 2)
frame_num += 1
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
#print("FPS: %.2f" % fps)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
# if output flag is set, save video file
if FLAGS.output_vid:
out.write(result)
#if cv2.waitKey(1) & 0xFF == ord('q'): break
print("Total Processing time: ", time.time() - start_process)
cv2.destroyAllWindows()
# saving the detections data into a csv
detections_df.to_csv(output_csv_path, index=False)
print("The detections file was successfully saved!")
# Adding the video counts data to the dataframe
if (camera == 1) or (camera == 2):
count_df_in = pd.DataFrame({
'Store_name': [store_name],
'Start_date': [time_start_vid_dt],
'End_date': [time_end_vid_dt],
'Camera': [camera],
'Count': [total_count_in],
'inout': "In",
'name_video': [file_name]
})
count_df_out = pd.DataFrame({
'Store_name': [store_name],
'Start_date': [time_start_vid_dt],
'End_date': [time_end_vid_dt],
'Camera': [camera],
'Count': [total_count_out],
'inout': "Out",
'name_video': [file_name]
})
else:
count_df_in = pd.DataFrame({
'Store_name': [store_name],
'Start_date': [time_start_vid_dt],
'End_date': [time_end_vid_dt],
'Camera': [camera],
'Count': [0],
'inout': "In",
'name_video': [file_name]
})
count_df_out = pd.DataFrame({
'Store_name': [store_name],
'Start_date': [time_start_vid_dt],
'End_date': [time_end_vid_dt],
'Camera': [camera],
'Count': [0],
'inout': "Out",
'name_video': [file_name]
})
count_df = pd.concat([count_df_in, count_df_out], ignore_index=True)
# saving the count data into into a csv
count_df.to_csv(count_csv_path, index=False)
print("The counts files were successfully saved!")
#upload the detections data to the database
upload_to_db(detections_df, 'tracker',
'append') # passing the dataframe
#upload the count data to the database
upload_to_db(count_df, 'counts', 'append')
def main(_argv):
with open("./config_birdview.yml", "r") as ymlfile:
bird_view_cfg = yaml.load(ymlfile)
width_og, height_og = 0, 0
corner_points = []
for section in bird_view_cfg:
corner_points.append(bird_view_cfg["image_parameters"]["p1"])
corner_points.append(bird_view_cfg["image_parameters"]["p2"])
corner_points.append(bird_view_cfg["image_parameters"]["p3"])
corner_points.append(bird_view_cfg["image_parameters"]["p4"])
width_og = int(bird_view_cfg["image_parameters"]["width_og"])
height_og = int(bird_view_cfg["image_parameters"]["height_og"])
img_path = bird_view_cfg["image_parameters"]["img_path"]
size_height = bird_view_cfg["image_parameters"]["size_height"]
size_width = bird_view_cfg["image_parameters"]["size_width"]
tr = np.array([
bird_view_cfg["image_parameters"]["p4"][0],
bird_view_cfg["image_parameters"]["p4"][1],
])
tl = np.array([
bird_view_cfg["image_parameters"]["p2"][0],
bird_view_cfg["image_parameters"]["p2"][1],
])
br = np.array([
bird_view_cfg["image_parameters"]["p3"][0],
bird_view_cfg["image_parameters"]["p3"][1],
])
bl = np.array([
bird_view_cfg["image_parameters"]["p1"][0],
bird_view_cfg["image_parameters"]["p1"][1],
])
widthA = np.sqrt(((br[0] - bl[0])**2) + ((br[1] - bl[1])**2))
widthB = np.sqrt(((tr[0] - tl[0])**2) + ((tr[1] - tl[1])**2))
maxWidth = max(int(widthA), int(widthB))
heightA = np.sqrt(((tr[0] - br[0])**2) + ((tr[1] - br[1])**2))
heightB = np.sqrt(((tl[0] - bl[0])**2) + ((tl[1] - bl[1])**2))
maxHeight = max(int(heightA), int(heightB))
matrix, imgOutput = compute_perspective_transform(corner_points, maxWidth,
maxHeight,
cv2.imread(img_path))
height, width, _ = imgOutput.shape
dim = (width, height)
# Definition of the parameters
max_cosine_distance = 0.4
nn_budget = None
nms_max_overlap = 1.0
# initialize deep sort
model_filename = "model_data/mars-small128.pb"
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# load tflite model if flag is set
if FLAGS.framework == "tflite":
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures["serving_default"]
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
output_video_1, output_video_2 = None, None
# get video ready to save locally if flag is set
if FLAGS.output:
# by default VideoCapture returns float instead of int
"""
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
"""
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
frame_num = 0
# while video is running
while True:
black_img = cv2.imread("./black_bg.png")
black_img = cv2.resize(black_img, dim, interpolation=cv2.INTER_AREA)
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print("Video has ended or failed, try a different video format!")
break
frame_num += 1
print("Frame #: ", frame_num)
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.0
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
# run detections on tflite if flag is set
if FLAGS.framework == "tflite":
interpreter.set_tensor(input_details[0]["index"], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]["index"])
for i in range(len(output_details))
]
# run detections using yolov3 if flag is set
if FLAGS.model == "yolov3" and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(
pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]),
)
else:
boxes, pred_conf = filter_boxes(
pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant([input_size, input_size]),
)
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
(
boxes,
scores,
classes,
valid_detections,
) = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score,
)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
# allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
allowed_classes = ["person"]
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(
frame,
"Objects being tracked: {}".format(count),
(5, 35),
cv2.FONT_HERSHEY_COMPLEX_SMALL,
2,
(0, 255, 0),
2,
)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
# initialize color map
cmap = plt.get_cmap("tab20b")
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes,
nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
bbox_array = []
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
bbox_array.append(
(int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3])))
class_name = track.get_class()
# draw bbox on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
cv2.rectangle(
frame,
(int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])),
color,
2,
)
cv2.rectangle(
frame,
(int(bbox[0]), int(bbox[1] - 30)),
(
int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1]),
),
color,
-1,
)
cv2.putText(
frame,
class_name + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)),
0,
0.75,
(255, 255, 255),
2,
)
# if enable info flag then print details about each track
if FLAGS.info:
print(
"Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}"
.format(
str(track.track_id),
class_name,
(int(bbox[0]), int(bbox[1]), int(bbox[2]), int(
bbox[3])),
))
if len(bbox_array) >= 1:
array_centroids, array_groundpoints = get_centroids_and_groundpoints(
bbox_array)
transformed_downoids = compute_point_perspective_transformation(
matrix, array_centroids)
# Show every point on the top view image
for point in transformed_downoids:
x, y = point
cv2.circle(black_img, (x, y), 60, (0, 255, 0), 2)
cv2.circle(black_img, (x, y), 3, (0, 255, 0), -1)
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(frame)
# result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
if output_video_1 is None and output_video_2 is None:
fourcc1 = cv2.VideoWriter_fourcc(*"MJPG")
output_video_1 = cv2.VideoWriter(
"./video.avi", fourcc1, 25,
(frame.shape[1], frame.shape[0]), True)
fourcc2 = cv2.VideoWriter_fourcc(*"MJPG")
output_video_2 = cv2.VideoWriter(
"./bird_view.avi",
fourcc2,
25,
(black_img.shape[1], black_img.shape[0]),
True,
)
elif output_video_1 is not None and output_video_2 is not None:
output_video_1.write(frame)
output_video_2.write(black_img)
if cv2.waitKey(1) & 0xFF == ord("q"):
break
cv2.destroyAllWindows()
def main(_argv):
# Definition of the parameters
max_cosine_distance = 0.4
nn_budget = None
nms_max_overlap = 1.0
# initialize deep sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = False
config.gpu_options.per_process_gpu_memory_fraction = 0.1
_ = InteractiveSession(config=config)
utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# load tflite model if flag is set
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(
model_path=f'{FLAGS.weights}_{FLAGS.size}')
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(
f'{FLAGS.weights}_{FLAGS.size}', tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
# get video ready to save locally if flag is set
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
all_start_time = None
frame_num = 0
# while video is running
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
Image.fromarray(frame)
else:
fps = float(frame_num) / (time.time() - all_start_time)
print("fps=%.2f size=%d frames=%d deep=%s output=%s" %
(fps, FLAGS.size, frame_num,
"true" if FLAGS.deep else "false", FLAGS.output))
break
frame_num += 1
if FLAGS.info:
print("frame_num=%d" % frame_num)
start_time = time.time()
if all_start_time is None:
all_start_time = time.time()
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
# run detections using yolov3 if flag is set
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for _, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(frame, "Objects being tracked: {}".format(count),
(5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2,
(0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
if FLAGS.deep:
features = encoder(frame, bboxes)
else:
features = np.empty((len(bboxes), 0), np.float32)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes,
nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
# draw bbox on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(frame, class_name + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
# if enable info flag then print details about each track
if FLAGS.info:
print(
"Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}"
.format(str(track.track_id), class_name, (int(
bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
out.write(result)
# calculate frames per second of running detections
if FLAGS.info:
fps = 1.0 / (time.time() - start_time)
print("fps=%.2f" % fps)
if not FLAGS.dont_show:
if cv2.waitKey(1) & 0xFF == ord('q'): break
if not FLAGS.dont_show:
cv2.destroyAllWindows()
def startRecording_YOLO():
date_and_time = time.strftime("%Y%m%d-%H-%M-%S") #Stores current date and time in YYYY-MM-DD-HH:MM format
vid_out_path = os.path.join(PROJECT_DIR, 'YoloV4', 'outputs', date_and_time + '.avi')
#vid = cv2.VideoCapture(test_drive) #0 for webcam/Raspberry Pi Cam
videothread = VideoThread(resolution=(640,480), framerate=30).start()
width = int(videothread.stream.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(videothread.stream.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(videothread.stream.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*'XVID')
output_video = cv2.VideoWriter(vid_out_path, codec, fps, (width,height))
#width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
#height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
#fps = int(vid.get(cv2.CAP_PROP_FPS))
#codec = cv2.VideoWriter_fourcc(*'XVID')
#output_video = cv2.VideoWriter(vid_out_path, codec, fps, (width,height))
frame_number = 0
freq = cv2.getTickFrequency()
avg_fps = 0
#while video is running/recording
while True:
return_val, frame = videothread.read()
#return_val, frame = vid.read()
if return_val:
#frame = cv2.flip(frame, -1)
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video error, try another format')
break
frame_number += 1
#print('Frame #: ', frame_number)
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data/ 255.
#mage_data = np.expand_dims(frame_resized, axis = 0)
#if floating_model:
# image_data = (np.float32(image_data) - 127.5)/127.5
image_data = image_data[np.newaxis, ...].astype(np.float32) #Converts image data to a float32 type
start_time = time.time()
#TFLite Detections
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
prediction = [interpreter.get_tensor(output_details[i]['index']) for i in range(len(output_details))]
#box = interpreter.get_tensor(output_details[0]['index'])[0]
#scores = interpreter.get_tensor(output_details[2]['index'])[0]
boxes, prediction_conf = filter_boxes(prediction[0], prediction[1], score_threshold=0.4, input_shape=tf.constant([input_size, input_size]))
#Reshape = returns a new tensor that has the same values as tensor in the same order, but with a new shape given by shape
#Shape = returns a 1-D integer tensor, represents the shape of the input
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes = tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores = tf.reshape(prediction_conf, (tf.shape(prediction_conf)[0], -1, tf.shape(prediction_conf)[-1])),
max_output_size_per_class = 50,
max_total_size = 50,
iou_threshold = 0.45,
score_threshold = 0.5
)
#convert the received data into numpy arrays, then slice out unused elements
number_of_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0 : int(number_of_objects)]
scores = scores.numpy()[0]
scores = scores[0 : int(number_of_objects)]
classes = classes.numpy()[0]
classes = classes[0 : int(number_of_objects)]
#format bounding boxes with normalized minimums and maximums of x and y
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
prediction_bbox = [bboxes, scores, classes, number_of_objects]
#Read in all the class names from config and only allow certain ones to be detected (eases computation power)
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
allowed_classes = ['traffic light', 'person', 'car', 'stop sign']
#loop through objects and get classification name, using only the ones allows in allowed_classes
names = []
deleted_indx = []
for i in range(number_of_objects):
classification_index = int(classes[i])
class_name = class_names[classification_index]
if class_name not in allowed_classes: deleted_indx.append(i)
else: names.append(class_name)
names = np.array(names)
count = len(names)
#delete irrelevant detections (not in allowed_classes)
bboxes = np.delete(bboxes, deleted_indx, axis = 0)
scores = np.delete(scores, deleted_indx, axis = 0)
#Feed tracker with encoded yolo detections
detections_features = encoder(frame, bboxes)
detections = [Detection(bbox, score, class_name, detection_feature) for bbox, score, class_name, detection_feature in zip(bboxes, scores, names, detections_features)]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
#run non-maxima supression (reduces amount of detected entities to as little as possible)
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
#Call tracker
tracker.predict()
tracker.update(detections)
#update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1: continue
bbox = track.to_tlbr()
class_name = track.get_class()
#if class_name == 'person': print('person found')
#change frame to that which showcases the lane detection
#frame = lane_detect.detect_edges(frame) #COMMENT OUT IF/WHEN ERROR OCCURS
#distance approximation (barebones, needs more adjusting)
cam_parameter = 18 #change with different cameras. Gets the detected distance closer to actual distance
distance = (np.pi)/(bbox[2].item() + bbox[3].item()) * 1000 + cam_parameter
det_dest = str(int(distance))
#draw bounded box on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1] - 30)), (int(bbox[0]) + (len(class_name) + len(det_dest)) * 18, int(bbox[1])), color, -1)
#cv2.putText(frame, class_name + "-" + str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)), 0, 0.75, (255, 255, 255), 2)
cv2.putText(frame, class_name + ": " + str(int(distance)), (int(bbox[0]), int(bbox[1] - 10)), 0, 0.75, (255, 255, 255), 2)
#calculate fps of running detections
fps = 1.0/ (time.time() - start_time)
avg_fps = avg_fps + fps
#print("FPS: %.2f" % fps)
cv2.putText(frame, "FPS: " + str(int(fps)), (width - 100, height - 20),0, 0.75, (255,255,255),2)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
cv2.imshow("Output Video", result)
output_video.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
cv2.destroyAllWindows()
print('Average FPS: ', (avg_fps/frame_number))
print('Number of Frames: ', frame_number)
videothread.stop()
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
images = FLAGS.images
# load model
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights, tags=[tag_constants.SERVING])
# loop through images in list and run Yolov4 model on each
for count, image_path in enumerate(images, 1):
original_image = cv2.imread(image_path)
original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
image_data = cv2.resize(original_image, (input_size, input_size))
image_data = image_data / 255.
images_data = []
for i in range(1):
images_data.append(image_data)
images_data = np.asarray(images_data).astype(np.float32)
if FLAGS.framework == 'tflite':
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
interpreter.set_tensor(input_details[0]['index'], images_data)
interpreter.invoke()
pred = [interpreter.get_tensor(output_details[i]['index']) for i in range(len(output_details))]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1], pred[0], score_threshold=0.25, input_shape=tf.constant([input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0], pred[1], score_threshold=0.25, input_shape=tf.constant([input_size, input_size]))
else:
infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score
)
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
pred_bbox = [bboxes, scores.numpy()[0], classes.numpy()[0], valid_detections.numpy()[0]]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
# allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to allow detections for only people)
# allowed_classes = ['person']
allowed_classes = ['Mask Person']
image = utils.draw_bbox(original_image, pred_bbox, allowed_classes = allowed_classes)
if FLAGS.covid:
distance = social_distance(pred_bbox,original_image, allowed_classes)
image = Image.fromarray(image.astype(np.uint8))
if not FLAGS.dont_show:
image.show()
image = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2RGB)
cv2.imwrite(FLAGS.output + 'detection' + str(count) + '.png', image)
def main(self, frame_data):
# Definition of the parameters
nms_max_overlap = 1.0
# set HyperParams
size = 416
iou = 0.45
score = 0.50
info = False
input_size = size
self.person1.is_used = 0
self.person2.is_used = 0
self.person3.is_used = 0
self.person4.is_used = 0
out = None
frame_data = cv2.cvtColor(frame_data, cv2.COLOR_BGR2RGB)
image_data = cv2.resize(frame_data, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
# start_time = time.time()
batch_data = tf.constant(image_data)
pred_bbox = self.infer(batch_data) # Yolo 모델 통과시켜서 바운딩 박스 좌표 반환
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4] # 좌표
pred_conf = value[:, :, 4:] # 벡터값
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=iou,
score_threshold=score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame_data.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
# allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if count:
cv2.putText(frame_data, "Objects being tracked: {}".format(count),
(5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2,
(0, 255, 0), 2)
# print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = self.encoder(frame_data, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
# initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes,
nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# DeepSort Tracking Start
# Call the tracker
self.tracker.predict() # load tracker
self.tracker.update(detections)
match_person = 0
# reset unmatched for center compare
unmatched = []
# update tracks
for track in self.tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
# draw bbox on screen # 이거 처리까지 하고 나서 보내야 할 것 같다.
bbox = track.to_tlbr()
class_name = track.get_class()
# Matching index with index_stack
if self.person1.is_exist(track.track_id):
self.person1.centerX, self.person1.centerY = self.getCenter(
bbox)
self.draw_box(frame_data, self.person1.index_stack[0], colors,
bbox)
self.person1.is_used = 1
match_person += 1
elif self.person2.is_exist(track.track_id):
self.person2.centerX, self.person2.centerY = self.getCenter(
bbox)
self.draw_box(frame_data, self.person2.index_stack[0], colors,
bbox)
self.person2.is_used = 1
match_person += 1
elif self.person3.is_exist(track.track_id):
self.person3.centerX, self.person3.centerY = self.getCenter(
bbox)
self.draw_box(frame_data, self.person3.index_stack[0], colors,
bbox)
self.person3.is_used = 1
match_person += 1
elif self.person4.is_exist(track.track_id):
self.person4.centerX, self.person4.centerY = self.getCenter(
bbox)
self.draw_box(frame_data, self.person4.index_stack[0], colors,
bbox)
self.person4.is_used = 1
match_person += 1
else:
unmatched.append([track.track_id, bbox])
print('found new object!')
unmatched = np.array(unmatched, dtype=object)
# Missed Person Only 1
if match_person == 3 and len(unmatched) == 1:
if self.person1.is_used == 0:
self.person1.centerX, self.person1.centerY = self.getCenter(
unmatched[0][1])
self.person1.index_stack.append(unmatched[0][0])
self.draw_box(frame_data, self.person1.index_stack[0], colors,
unmatched[0][1])
self.person1.is_used = 1
match_person += 1
elif self.person2.is_used == 0:
self.person2.centerX, self.person2.centerY = self.getCenter(
unmatched[0][1])
self.person2.index_stack.append(unmatched[0][0])
self.draw_box(frame_data, self.person2.index_stack[0], colors,
unmatched[0][1])
self.person2.is_used = 1
match_person += 1
elif self.person3.is_used == 0:
self.person3.centerX, self.person3.centerY = self.getCenter(
unmatched[0][1])
self.person3.index_stack.append(unmatched[0][0])
self.draw_box(frame_data, self.person3.index_stack[0], colors,
unmatched[0][1])
self.person3.is_used = 1
match_person += 1
elif self.person4.is_used == 0:
self.person4.centerX, self.person4.centerY = self.getCenter(
unmatched[0][1])
self.person4.index_stack.append(unmatched[0][0])
self.draw_box(frame_data, self.person4.index_stack[0], colors,
unmatched[0][1])
self.person4.is_used = 1
match_person += 1
else:
print("ERROR : Something problem on object.is_used")
# Missed Person Over 2
if match_person <= 3 and len(unmatched) >= 1:
for unmatch in unmatched:
if match_person >= 4:
break
else:
# Apply center location Euclidean Distance
EUD_min = self.get_EuclideanDistance(unmatch)
print(EUD_min)
if not len(str(EUD_min)) == 0:
self.draw_box(frame_data, EUD_min, colors, unmatch[1])
match_person += 1
# if enable info flag then print details about each track
if info:
print(
"Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}"
.format(
str(track.track_id), class_name,
(int(bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
result = cv2.cvtColor(frame_data, cv2.COLOR_RGB2BGR)
return result
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
images = FLAGS.images
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
# loop through images in list and run Yolov4 model on each
for count, image_path in enumerate(images, 1):
original_image = cv2.imread(image_path)
original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
image_data = cv2.resize(original_image, (input_size, input_size))
image_data = image_data / 255.
# get image name by using split method
image_name = image_path.split('/')[-1]
image_name = image_name.split('.')[0]
images_data = []
for i in range(1):
images_data.append(image_data)
images_data = np.asarray(images_data).astype(np.float32)
infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
# run non max suppression on detections
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=0.5,
score_threshold=0.5)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
# hold all detection data in one variable
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
crop_path = os.path.join(os.getcwd(), 'detections', 'crop')
try:
os.mkdir(crop_path)
except FileExistsError:
pass
crop_objects(cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB),
pred_bbox, crop_path, allowed_classes)
def main(_argv):
# Definition of the parameters
max_cosine_distance = 0.4
nn_budget = None
nms_max_overlap = 1.0
# initialize deep sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
video_path = FLAGS.video
# load tflite model if flag is set
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
# get video ready to save locally if flag is set
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
frame_num = 0
# while video is running
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_num += 1
print('Frame #: ', frame_num)
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
# run detections using yolov3 if flag is set
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
#allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(frame, "Objects being tracked: {}".format(count),
(5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2,
(0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
# initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes,
nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
# draw bbox on screen
# names = {'6_d': 'Thomas Delaney',
# '10_b': 'Leroy Sane',
# '18_b': 'Leon Goretzka',
# '25_b': 'Thomas Muller',
# '5_d': 'Dan-Axel Zagadou',
# '12_d': 'Zaragoza',
# '4_b': 'Niklas Sule',
# '14_d': 'Nico Schulz',
# '11_d': 'Marco Reus',
# 'Referee': 'Referee',
# 'ball': 'ball',
# '10_d': 'Thorgan Hazard',
# '6_b': 'Joshua Kimmich ',
# 'gk_b': 'Ron-Thorben Hoffmann(GK)',
# '17_b': 'Jérôme Boateng',
# '27_b': 'David Alaba',
# '9_d': 'Erling Haaland',
# '8_d': 'Mahmoud Dahoud',
# 'gk_d': 'Luca Unbehaun(GK)',
# '19_b': 'Alphonso Davies',
# '29_b': 'Kingsley Coman',
# '24_d': 'Marcel Schmelzer',
# '9_b': 'Robert Lewandowski',
# "23_d": 'Emre Can',
# }
# if class_name == 'Referee':
# color = (0, 0, 0)
if class_name == 'ball':
# color = (255, 255, 255)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 1)
# else:
# try:
# colors = {'b': (252, 3, 78), 'd': (250, 247, 80)}
# color = colors[str(class_name.split('_')[-1])]
# except KeyError:
# pass
# class_name = names[str(class_name)]
# color = (250, 247, 80)
# color = colors[int(track.track_id) % len(colors)]
# color = [i * 255 for i in color]
# cv2.rectangle(frame, (int(bbox[0]), int(
# bbox[1])), (int(bbox[2]), int(bbox[3])), color, 1)
# cv2.rectangle(frame, (int(bbox[0]), int(
# bbox[1]-30)), (int(bbox[0])+(len(str(class_name)))*17, int(bbox[1])), color, -1)
cv2.putText(frame, class_name, (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 251, 46), 2)
# if enable info flag then print details about each track
if FLAGS.info:
print(
"Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}"
.format(str(track.track_id), class_name, (int(
bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cv2.destroyAllWindows()
def main(_argv):
# Definition of the parameters
max_cosine_distance = 0.4
nn_budget = None
nms_max_overlap = 1.0
# initialize deep sort
model_filename = cfg.PATH + '/model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
# tf
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
while True:
data = sys.stdin.readline()
if data:
data = json.loads(data)
if data['end']:
break
frame = np.array(data['frame_image'], dtype=np.uint8)
image_data = frame / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
# tf
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
#allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(
boxs, classes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# ds = []
# for detection in detections:
# d = dict()
# d["bbox"] = detection.tlwh.tolist()
# d["confidence"] = detection.confidence
# d["class"] = detection.class_name
# ds.append(d)
#
# # send data to Node (without tracking...)
# print(json.dumps(ds))
#Call the tracker
tracker.predict()
tracker.update(detections)
# Store tracks for json...
tracks = []
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
class_name = track.get_class()
t = dict()
bbs = track.to_tlbr().tolist()
t["class"] = class_name
bbox = dict()
bbox["left"] = bbs[0]
bbox["top"] = bbs[1]
bbox["right"] = bbs[2]
bbox["bottom"] = bbs[3]
t["bbox"] = bbox
t["id"] = track.track_id
t["score"] = track.detection_actual_score
tracks.append(t)
#send data to Node!
print(json.dumps(tracks))
def main(_argv):
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
images = FLAGS.images
# load model
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
# loop through images in list and run Yolov4 model on each
for count, image_path in enumerate(images, 1):
original_image = cv2.imread(image_path)
original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
image_data = cv2.resize(original_image, (input_size, input_size))
image_data = image_data / 255.
# get image name by using split method
image_name = image_path.split('/')[-1]
image_name = image_name.split('.')[0]
images_data = []
for i in range(1):
images_data.append(image_data)
images_data = np.asarray(images_data).astype(np.float32)
if FLAGS.framework == 'tflite':
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
interpreter.set_tensor(input_details[0]['index'], images_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
infer = saved_model_loaded.signatures['serving_default']
batch_data = tf.constant(images_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
# run non max suppression on detections
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = original_image.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h, original_w)
# hold all detection data in one variable
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to allow detections for only people)
#allowed_classes = ['person']
# if crop flag is enabled, crop each detection and save it as new image
if FLAGS.crop:
crop_path = os.path.join(os.getcwd(), 'detections', 'crop',
image_name)
try:
os.mkdir(crop_path)
except FileExistsError:
pass
crop_objects(cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB),
pred_bbox, crop_path, allowed_classes)
# if ocr flag is enabled, perform general text extraction using Tesseract OCR on object detection bounding box
if FLAGS.ocr:
ocr(cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB), pred_bbox)
# if count flag is enabled, perform counting of objects
if FLAGS.count:
# count objects found
counted_classes = count_objects(pred_bbox,
by_class=False,
allowed_classes=allowed_classes)
# loop through dict and print
for key, value in counted_classes.items():
print("Number of {}s: {}".format(key, value))
image = utils.draw_bbox(original_image,
pred_bbox,
FLAGS.info,
counted_classes,
allowed_classes=allowed_classes,
read_plate=FLAGS.plate)
else:
image = utils.draw_bbox(original_image,
pred_bbox,
FLAGS.info,
allowed_classes=allowed_classes,
read_plate=FLAGS.plate)
image = Image.fromarray(image.astype(np.uint8))
if not FLAGS.dont_show:
image.show()
image = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2RGB)
cv2.imwrite(FLAGS.output + 'detection' + str(count) + '.png', image)
def main(self, frame_data):
# Definition of the parameters
nms_max_overlap = 1.0
# set HyperParams
size = 416
iou = 0.45
score = 0.50
info = False
people_num = 4
input_size = size
self.indexing.queue.clear()
for k in range(people_num):
self.indexing.put(k+1)
out = None
frame_data = cv2.cvtColor(frame_data, cv2.COLOR_BGR2RGB)
image_data = cv2.resize(frame_data, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
batch_data = tf.constant(image_data)
pred_bbox = self.infer(batch_data) # Yolo 모델 통과시켜서 바운딩 박스 좌표 반환
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4] # 좌표
pred_conf = value[:, :, 4:] # 벡터값
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf, (tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=iou,
score_threshold=score
)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame_data.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
# allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if count:
cv2.putText(frame_data, "Objects being tracked: {}".format(count), (5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL,
2,(0, 255, 0), 2)
# print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = self.encoder(frame_data, bboxes)
detections = [Detection(bbox, score, class_name, feature) for bbox, score, class_name, feature in
zip(bboxes, scores, names, features)]
# initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# DeepSort Tracking Start
# Call the tracker
self.tracker.predict() # load tracker
self.tracker.update(detections)
#check is_confirmed
is_not_confirmed = 0
tracks_count = 0
for w, track in enumerate(self.tracker.tracks):
tracks_count += 1
# print('count', tracks_count)
# update tracks
for index, track in enumerate(self.tracker.tracks):
if not track.is_confirmed() or track.time_since_update > 1:
is_not_confirmed += 1
continue
if index-is_not_confirmed+1 > people_num:
break
bbox = track.to_tlbr()
class_name = track.get_class()
# draw bbox on screen # 이거 처리까지 하고 나서 보내야 할 것 같다.
for i in range(self.indexing.qsize()):
check_index = self.indexing.get()
if track.track_id == check_index:
color = colors[int(track.track_id)*8 % len(colors)]
color = [j * 255 for j in color]
cv2.rectangle(frame_data, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(frame_data, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) + (len(class_name) + len(str(track.track_id))) * 17, int(bbox[1])),
color, -1)
cv2.putText(frame_data, class_name + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
break
else:
self.indexing.put(check_index)
if i == self.indexing.qsize() - 1:
cng_index = self.indexing.get()
print('index changed', track.track_id, '->', cng_index)
# track.track_id = cng_index
color = colors[int(cng_index)*8 % len(colors)]
color = [j * 255 for j in color]
cv2.rectangle(frame_data, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(frame_data, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) + (len(class_name) + len(str(cng_index))) * 17, int(bbox[1])),
color, -1)
cv2.putText(frame_data, class_name + "-" + str(cng_index),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
# if enable info flag then print details about each track
if info:
print("Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}".format(str(track.track_id),
class_name, (
int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3]))))
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
# print("FPS: %.2f" % fps)
result = cv2.cvtColor(frame_data, cv2.COLOR_RGB2BGR)
return result
def _capture_loop(self):
dt = 1 / self.fps
self.FLAGS = self.FLAGS()
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(self.FLAGS)
input_size = self.FLAGS.size
saved_model_loaded = tf.saved_model.load(self.FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
logger.debug("Observation started")
out = None
while self.isrunning:
return_value, frame = self.vid.read()
if return_value:
if len(self.frames) == self.max_frames:
self.frames = self.frames[1:]
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=self.FLAGS.iou,
score_threshold=self.FLAGS.score)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, xmax, ymax
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(boxes.numpy()[0], original_h,
original_w)
pred_bbox = [
bboxes,
scores.numpy()[0],
classes.numpy()[0],
valid_detections.numpy()[0]
]
image = utils.draw_bbox(frame, pred_bbox, self.FLAGS.info)
self.frames.append(frame)
time.sleep(dt)
logger.info("Thread stopped successfully")
def main(_argv):
# Definition of the parameters
nms_max_overlap = 1.0
# initialize deep sort parameters
encoder, tracker = init_deepsort_params()
# load configuration for object detector
input_size, video_path = load_obj_detector_cfg()
# load tflite model if flag is set
if FLAGS.framework == 'tflite':
tfl = tf_lite_ngine()
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
vid = get_video_stream(video_path)
# get video ready to save locally if flag is set
out = None
if FLAGS.output:
out = init_video_out(vid)
frame_num = 0
# while video is running
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_num += 1
print('Frame #: ', frame_num)
frame_size = frame.shape[:2]
image_data = init_image_data(frame, input_size)
start_time = time.time()
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
boxes, pred_conf = tfl.detect(image_data)
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = apply_tf_nms(
boxes, pred_conf)
# convert data to numpy arrays and slice out unused elements
num_objects, bboxes, scores, classes = detections_to_np_array(
valid_detections, boxes, scores, classes)
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# fetch allowed object classes, ignore the rest classes
names, deleted_indx = get_allowed_obj_classes(classes, num_objects)
if FLAGS.count:
show_tracked_object_count(names, frame)
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
# process detections with YOLO tracker
frame = process_detections(tracker, detections, nms_max_overlap, frame)
# calculate and print frames per second of running detections
print_fps(start_time, time.time(), frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
cv2.destroyAllWindows()
def main(_argv):
# Definition of the parameters
max_cosine_distance = 0.4
nn_budget = None
nms_max_overlap = 1.0
# initialize deep sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
STRIDES, ANCHORS, NUM_CLASS, XYSCALE = utils.load_config(FLAGS)
input_size = FLAGS.size
#images = FLAGS.images
video_path = FLAGS.video
# load tflite model if flag is set
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
# begin video capture
try:
vid = cv2.VideoCapture(int(video_path))
except:
vid = cv2.VideoCapture(video_path)
out = None
# get video ready to save locally if flag is set
if FLAGS.output:
# by default VideoCapture returns float instead of int
width = int(vid.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(vid.get(cv2.CAP_PROP_FPS))
codec = cv2.VideoWriter_fourcc(*FLAGS.output_format)
out = cv2.VideoWriter(FLAGS.output, codec, fps, (width, height))
frame_num = 0
if FLAGS.shirt:
allowed_classes = ['Shirt']
#ROI =
if FLAGS.trouser:
allowed_classes = ['Trousers']
if FLAGS.jeans:
allowed_classes = ['Jeans']
if FLAGS.dress:
allowed_classes = ['Dress']
if FLAGS.footwear:
allowed_classes = ['Footwear']
if FLAGS.jacket:
allowed_classes = ['Jacket']
if FLAGS.skirt:
allowed_classes = ['Skirt']
if FLAGS.suit:
allowed_classes = ['Suit']
# while video is running
while True:
return_value, frame = vid.read()
if return_value:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(frame)
else:
print('Video has ended or failed, try a different video format!')
break
frame_num += 1
print('Frame #: ', frame_num)
frame_size = frame.shape[:2]
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
# run detections using yolov3 if flag is set
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for key, value in pred_bbox.items():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=50,
max_total_size=50,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
#allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(frame, "Objects being tracked: {}".format(count),
(5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2,
(0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes,
nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
# draw bbox on screen
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
#cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), color, 2)
#cv2.rectangle(frame, (int(bbox[0]), int(bbox[1]-30)), (int(bbox[0])+(len(class_name)+len(str(track.track_id)))*17, int(bbox[1])), color, -1)
#cv2.putText(frame, class_name + "-" + str(track.track_id),(int(bbox[0]), int(bbox[1]-10)),0, 0.75, (255,255,255),2)
# if enable info flag then print details about each track
if FLAGS.info:
print(
"Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}"
.format(str(track.track_id), class_name, (int(
bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
if FLAGS.color:
PATH = './training.data'
#(int(bbox[0])):(int(bbox[2])), (int(bbox[1])):(int(bbox[3]))
#ROI = frame[(int(bbox[0]) +50) :(int(bbox[2]) - 50), (int(bbox[1])+ 50):(int(bbox[3])-50)]
#ROI = frame[(int(bbox[1])) +15 :(int(bbox[3])-15),(int(bbox[0])+15):(int(bbox[2])-15)]
ROI = frame[int((int(bbox[1]) + int(bbox[3])) /
2):int((int(bbox[1]) + int(bbox[3])) / 2) + 1,
int((int(bbox[0]) + int(bbox[2])) /
2):int((int(bbox[0]) + int(bbox[2])) / 2) + 1]
#ROI = frame[(int(bbox[1])):(int(bbox[3])),(int(bbox[0])):(int(bbox[2]))]
#ROI = frame[int(0.5* (int(bbox[1] - 50)+ int(bbox[3] + 50))),int(0.5*(int(bbox[0] - 50) +int(bbox[2] + 50 )))]
#print(ROI)
color_histogram_feature_extraction.color_histogram_of_test_image(
ROI)
prediction = knn_classifier.main('training.data', 'test.data')
#prediction = 'red'
red = load_red('test.data')
Red = str(red)
#Red = str(Red_1)
print('this is the variable of the red:- ' + str(Red))
green = load_green('test.data')
Green = str(green)
#Green = str(Green_1)
print('this is the variable of the green:- ' + str(Green))
blue = load_blue('test.data')
#Blue_1 = int(blue)
Blue = str(blue)
print('this is the variable of the blue:- ' + str(Blue))
#hsv = rgb_to_hsv(red,green,blue)
#print("HSV: " + str(hsv))
if red and blue and green != None:
HLS = colorsys.rgb_to_hls(red, green, blue)
HUE = int(HLS[0])
Light = int(HLS[1])
Saturation = int(HLS[2])
print("HLS is equal to", HLS)
print('HUE: ', HUE)
print('LIGHT: ', Light)
print('Saturation', Saturation)
if red and blue and green != None:
HSV = rgb_to_hsv(red, green, blue)
HUE_1 = int(HSV[0])
Saturation_1 = int(HSV[1])
Value = int(HSV[2])
print("HSV is equal to", HSV)
print('Hue: ', HUE_1)
print('saturation: ', Saturation_1)
print('value', Value)
print(str(prediction) + " " + str(class_name))
if FLAGS.Fuzzy_black:
#if str(59.7) <= Red < str(200.9) and str(74) <= Blue < str(207) and str(70) <= Green < str(203):
if 0 <= HUE_1 < 210 and 0 <= Saturation_1 < 41 and 0 <= Value < 86:
print("THIS IS THE black COLOR yaaaaaaaaaaaaaaaaaaaa")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "BLACK" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_red:
#if str(139) <= Red < str(255) and str(0) <= Green < str(160) and str(0) <= Blue < str(128):
if 0 <= HUE_1 < 348 and 47 <= Saturation_1 < 100 and 55 <= Value < 100:
print(
"THIS IS THE red COLOR redddddddddddddddddddddddddddddddddddd"
)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "RED" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_orange:
#if str(255) <= Red < str(255) and str(69) <= Green < str(165) and str(0) <= Blue < str(80):
if 9 <= HUE_1 < 39 and 69 <= Saturation_1 < 100 and Value == 100:
print(
"THIS IS THE ORANGE COLOR orangeeeeeeeeeeeeeeeeeeeeeeee"
)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + "ORANGE" + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.Fuzzy_yellow:
#if str(189) <= Red < str(255) and str(183) <= Green < str(255) and str(0) <= Blue < str(224):
if 0 <= HUE_1 < 56 and 12 <= Saturation_1 < 100 and 74 <= Value < 100:
print("THIS IS THE YELLOW COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + "YELLOW" + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.Fuzzy_blue:
#if str(0) <= Red < str(176) and str(0) <= Green < str(244) and str(112) <= Blue < str(255):
if 187 <= HUE_1 < 240 and 21 <= Saturation_1 < 100 and 44 <= Value < 100:
print("THIS IS THE BLUE COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "BLUE" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_white:
#if str(240) <= Red < str(255) and str(228) <= Green < str(255) and str(215) <= Blue < str(255):
if 0 <= HUE_1 < 340 and 0 <= Saturation_1 < 14 and 96 <= Value < 100:
print("THIS IS THE WHITE COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "WHITE" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_purple:
#if str(72) <= Red < str(255) and str(0) <= Green < str(230) and str(128) <= Blue < str(255):
if 0 <= HUE_1 < 302 and 8 <= Saturation_1 < 100 and 50 <= Value < 100:
print("THIS IS THE PURPLE COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + "PURPLE" + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.Fuzzy_green:
#if str(0) <= Red < str(173) and str(100) <= Green < str(255) and str(0) <= Blue < str(170):
if 0 <= HUE_1 < 160 and 24 <= Saturation_1 < 100 and 39 <= Value < 100:
print("THIS IS THE green COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "GREEN" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_brown:
#if str(128) <= Red < str(255) and str(0) <= Green < str(248) and str(0) <= Blue < str(288):
if 0 <= HUE_1 < 48 and 14 <= Saturation_1 < 100 and 50 <= Value < 100:
print("THIS IS THE BROWN COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "BROWN" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_cyan:
#if str(0) <= Red < str(244) and str(128) <= Green < str(255) and str(128) <= Blue < str(255):
if 0 <= HUE_1 < 182 and 12 <= Saturation_1 < 100 and 50 <= Value < 100:
print("THIS IS THE CYAN COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "CYAN" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.Fuzzy_pink:
#if str(199) <= Red < str(255) and str(20) <= Green < str(192) and str(133) <= Blue < str(203):
if 322 <= HUE_1 < 351 and 25 <= Saturation_1 < 92 and 78 <= Value < 100:
print("THIS IS THE PINK COLOR")
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame,
class_name + " " + "PINK" + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
if FLAGS.black:
if prediction == 'black':
#ROI = frame[int((int(bbox[1]) + int(bbox[3]))/2):int((int(bbox[1]) + int(bbox[3]))/2)+1,int((int(bbox[0]) + int(bbox[2]))/2):int((int(bbox[0]) + int(bbox[2]))/2)+1]
#color_histogram_feature_extraction.color_histogram_of_test_image(ROI)
#prediction = knn_classifier.main('training.data','test.data')
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.blue:
if prediction == 'blue':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.red:
if prediction == 'red':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.yellow:
if prediction == 'yellow':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.orange:
if prediction == 'orange':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.violet:
if prediction == 'violet':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.white:
if prediction == 'white':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
if FLAGS.green:
if prediction == 'green':
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(
frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(
frame, class_name + " " + str(prediction) + "-" +
str(track.track_id), (int(bbox[0]), int(bbox[1] - 10)),
0, 0.75, (255, 255, 255), 2)
#cv2.putText(frame, class_name + " " + str(prediction) + "-" + str(track.track_id),(int(bbox[0]), int(bbox[1]-10)),0, 0.75, (255,255,255),2)
#print('ferture data:' +" " + feature_data)
#result_1 = np.asarray(frame)
#result_1 = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
#cv2.imshow('color classifier', result_1)
#print(color_histogram_feature_extraction.feature_data)
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# if output flag is set, save video file
if FLAGS.output:
out.write(result)
if cv2.waitKey(1) & 0xFF == ord('q'): break
cv2.destroyAllWindows()
def main(_argv):
global lock, buffer, bts
# Definition of the parameters
max_cosine_distance = 0.5
nn_budget = None
nms_max_overlap = 0.8
counter = []
# initialize deep sort
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
# calculate cosine distance metric
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# initialize tracker
tracker = Tracker(metric)
# load configuration for object detector
config = ConfigProto()
config.gpu_options.allow_growth = True
input_size = FLAGS.size
# load tflite model if flag is set
if FLAGS.framework == 'tflite':
interpreter = tf.lite.Interpreter(model_path=FLAGS.weights)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
print(input_details)
print(output_details)
# otherwise load standard tensorflow saved model
else:
saved_model_loaded = tf.saved_model.load(FLAGS.weights,
tags=[tag_constants.SERVING])
infer = saved_model_loaded.signatures['serving_default']
frame_num = 0
# while video is running
while True:
with lock:
bts += buffer
buffer = b''
time.sleep(0.01)
# jpghead = bts.find(b'\xff\xd8')
# jpgend = bts.find(b'\xff\xd9')
# if jpghead < 0 or jpgend < 0:
# continue
# if jpgend < jpghead:
# # raise Exception("{}..{}".format(jpghead, jpgend))
# continue
# print(jpghead, jpgend)
# jpg=bts[jpghead:jpgend+2]
# bts=bts[jpgend+2:]
jpghead = bts.find(b'\xff\xd8')
if jpghead >= 0:
bts = bts[jpghead:]
jpgend = bts.find(b'\xff\xd9')
if jpghead < 0 or jpgend < 0:
continue
print(jpghead, jpgend)
jpg = bts[0:jpgend + 2]
bts = bts[jpgend + 2:]
img = cv2.imdecode(np.frombuffer(jpg, dtype=np.uint8),
cv2.IMREAD_UNCHANGED)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
frame = cv2.resize(img, (640, 480))
cv2.imshow('input', frame)
cv2.waitKey(1)
frame_num += 1
print('Frame #: ', frame_num)
image_data = cv2.resize(frame, (input_size, input_size))
image_data = image_data / 255.
image_data = image_data[np.newaxis, ...].astype(np.float32)
start_time = time.time()
bts = bts[-6000:]
# run detections on tflite if flag is set
if FLAGS.framework == 'tflite':
interpreter.set_tensor(input_details[0]['index'], image_data)
interpreter.invoke()
pred = [
interpreter.get_tensor(output_details[i]['index'])
for i in range(len(output_details))
]
# run detections using yolov3 if flag is set
if FLAGS.model == 'yolov3' and FLAGS.tiny == True:
boxes, pred_conf = filter_boxes(pred[1],
pred[0],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
boxes, pred_conf = filter_boxes(pred[0],
pred[1],
score_threshold=0.25,
input_shape=tf.constant(
[input_size, input_size]))
else:
batch_data = tf.constant(image_data)
pred_bbox = infer(batch_data)
for value in pred_bbox.values():
boxes = value[:, :, 0:4]
pred_conf = value[:, :, 4:]
boxes, scores, classes, valid_detections = tf.image.combined_non_max_suppression(
boxes=tf.reshape(boxes, (tf.shape(boxes)[0], -1, 1, 4)),
scores=tf.reshape(
pred_conf,
(tf.shape(pred_conf)[0], -1, tf.shape(pred_conf)[-1])),
max_output_size_per_class=15,
max_total_size=15,
iou_threshold=FLAGS.iou,
score_threshold=FLAGS.score)
# convert data to numpy arrays and slice out unused elements
num_objects = valid_detections.numpy()[0]
bboxes = boxes.numpy()[0]
bboxes = bboxes[0:int(num_objects)]
scores = scores.numpy()[0]
scores = scores[0:int(num_objects)]
classes = classes.numpy()[0]
classes = classes[0:int(num_objects)]
# format bounding boxes from normalized ymin, xmin, ymax, xmax ---> xmin, ymin, width, height
original_h, original_w, _ = frame.shape
bboxes = utils.format_boxes(bboxes, original_h, original_w)
# store all predictions in one parameter for simplicity when calling functions
pred_bbox = [bboxes, scores, classes, num_objects]
# read in all class names from config
class_names = utils.read_class_names(cfg.YOLO.CLASSES)
# by default allow all classes in .names file
allowed_classes = list(class_names.values())
# custom allowed classes (uncomment line below to customize tracker for only people)
#allowed_classes = ['person']
# loop through objects and use class index to get class name, allow only classes in allowed_classes list
names = []
deleted_indx = []
for i in range(num_objects):
class_indx = int(classes[i])
class_name = class_names[class_indx]
if class_name not in allowed_classes:
deleted_indx.append(i)
else:
names.append(class_name)
names = np.array(names)
count = len(names)
if FLAGS.count:
cv2.putText(frame, "Objects being tracked: {}".format(count),
(5, 35), cv2.FONT_HERSHEY_COMPLEX_SMALL, 2,
(0, 255, 0), 2)
print("Objects being tracked: {}".format(count))
# delete detections that are not in allowed_classes
bboxes = np.delete(bboxes, deleted_indx, axis=0)
scores = np.delete(scores, deleted_indx, axis=0)
# encode yolo detections and feed to tracker
features = encoder(frame, bboxes)
detections = [
Detection(bbox, score, class_name, feature)
for bbox, score, class_name, feature in zip(
bboxes, scores, names, features)
]
#initialize color map
cmap = plt.get_cmap('tab20b')
colors = [cmap(i)[:3] for i in np.linspace(0, 1, 20)]
# run non-maxima supression
boxs = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.class_name for d in detections])
indices = preprocessing.non_max_suppression(boxs, classes,
nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
current_count = int(0)
# update tracks
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
class_name = track.get_class()
color = colors[int(track.track_id) % len(colors)]
color = [i * 255 for i in color]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), color, 2)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1] - 30)),
(int(bbox[0]) +
(len(class_name) + len(str(track.track_id))) * 17,
int(bbox[1])), color, -1)
cv2.putText(frame, class_name + "-" + str(track.track_id),
(int(bbox[0]), int(bbox[1] - 10)), 0, 0.75,
(255, 255, 255), 2)
center = (int(
((bbox[0]) + (bbox[2])) / 2), int(((bbox[1]) + (bbox[3])) / 2))
pts[track.track_id].append(center)
for j in range(1, len(pts[track.track_id])):
if pts[track.track_id][j - 1] is None or pts[
track.track_id][j] is None:
continue
thickness = int(np.sqrt(64 / float(j + 1)) * 2)
cv2.line(frame, (pts[track.track_id][j - 1]),
(pts[track.track_id][j]), color, thickness)
height, width, _ = frame.shape
cv2.line(frame, (0, int(3 * height / 6 + height / 2)),
(width, int(3 * height / 6 + height / 2)), (0, 255, 0),
thickness=2)
cv2.line(frame, (0, int(3 * height / 6 - height / 2)),
(width, int(3 * height / 6 - height / 2)), (0, 255, 0),
thickness=2)
center_y = int(((bbox[1]) + (bbox[3])) / 2)
if center_y <= int(3 * height / 6 +
height / 2) and center_y >= int(3 * height / 6 -
height / 2):
if class_name == 'Among_Us_Alive' or class_name == 'Among_Us_Dead':
counter.append(int(track.track_id))
current_count += 1
# if enable info flag then print details about each track
if FLAGS.info:
print(
"Tracker ID: {}, Class: {}, BBox Coords (xmin, ymin, xmax, ymax): {}"
.format(str(track.track_id), class_name, (int(
bbox[0]), int(bbox[1]), int(bbox[2]), int(bbox[3]))))
total_count = len(set(counter))
cv2.putText(frame, "Current Figurine Count: " + str(current_count),
(0, 80), 0, 1, (0, 0, 255), 2)
cv2.putText(frame, "Total Figurine Count: " + str(total_count),
(0, 130), 0, 1, (0, 0, 255), 2)
# calculate frames per second of running detections
fps = 1.0 / (time.time() - start_time)
print("FPS: %.2f" % fps)
result = np.asarray(frame)
result = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
if not FLAGS.dont_show:
cv2.imshow("Output Video", result)
# with lock:
# bts = bts[:5000]
# time.sleep(0.01)
if cv2.waitKey(1) & 0xFF == ord('q'): break
