def main():
input_video_filepath = sys.argv[1]
write_video_flag = False
output_video_filepath = ""
if len(sys.argv) > 2:
output_video_filepath = sys.argv[2]
write_video_flag = True
async_video_flag = False
config = OmegaConf.load("config.yaml")
detector = LineCrossingDetector(config)
counters = [PeopleCounter(**c) for c in config.people_counter]
if async_video_flag:
video_capture = VideoCaptureAsync(input_video_filepath)
else:
video_capture = cv2.VideoCapture(input_video_filepath)
if async_video_flag:
video_capture.start()
if write_video_flag:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*"XVID")
output_writer = cv2.VideoWriter(output_video_filepath, fourcc, 30,
(w, h))
frame_index = 0
pbar = tqdm(total=int(video_capture.get(cv2.CAP_PROP_FRAME_COUNT)))
while True and frame_index < 12000:
ret, frame = video_capture.read()
frame_index = frame_index + 1
if not ret:
break
if frame_index < 10000:
continue
detections = detector.detect(frame, visualize=True)
for counter in counters:
counter.update(detections, frame_index)
counter.visualize(frame)
for d in detections:
print(
f"Frame: {frame_index}. Track id: {d.track_id}. Line id: {d.line_id}"
)
if write_video_flag:
output_writer.write(frame)
pbar.update()
if async_video_flag:
video_capture.stop()
else:
video_capture.release()
if write_video_flag:
output_writer.release()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 1.0
output_format = 'mp4'
initialize_door_by_yourself = False
door_array = None
# Deep SORT
model_filename = '../model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
show_detections = True
writeVideo_flag = True
asyncVideo_flag = False
error_values = []
check_gpu()
files = sorted(os.listdir('data_files/videos'))
for video_name in files:
print("opening video: {}".format(video_name))
file_path = join('data_files/videos', video_name)
output_name = 'save_data/out_' + video_name[0:-3] + output_format
counter = Counter(counter_in=0, counter_out=0, track_id=0)
truth = get_truth(video_name)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(output_name, fourcc, 15, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
all_doors = read_door_info('data_files/doors_info_links.json')
door_array = all_doors[video_name]
rect_door = Rectangle(door_array[0], door_array[1], door_array[2],
door_array[3])
border_door = door_array[3]
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
y1 = (counter.counter_in - truth.inside)**2
y2 = (counter.counter_out - truth.outside)**2
total_count = counter.return_total_count()
true_total = truth.inside + truth.outside
if true_total != 0:
err = abs(total_count - true_total) / true_total
else:
err = abs(total_count - true_total)
mse = (y1 + y2) / 2
log_res = "in video: {}\n predicted / true\n counter in: {} / {}\n counter out: {} / {}\n" \
" total: {} / {}\n error: {}\n mse error: {}\n______________\n".format(video_name,
counter.counter_in,
truth.inside,
counter.counter_out,
truth.outside,
total_count,
true_total, err, mse)
with open('../log_results.txt', 'a') as log:
log.write(log_res)
print(log_res)
error_values.append(err)
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
cv2.rectangle(frame, (int(door_array[0]), int(door_array[1])),
(int(door_array[2]), int(door_array[3])),
(23, 158, 21), 3)
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
score = "%.2f" % (det.confidence * 100) + "%"
# rect_head = Rectangle(bbox[0], bbox[1], bbox[2], bbox[3])
# rect_door = Rectangle( int(door_array[0]), int(door_array[1]), int(door_array[2]), int(door_array[3]) )
# intersection = rect_head & rect_door
#
# if intersection:
# squares_coeff = rect_square(*intersection)/ rect_square(*rect_head)
# cv2.putText(frame, score + " inter: " + str(round(squares_coeff, 3)), (int(bbox[0]), int(bbox[3])), 0,
# 1e-3 * frame.shape[0], (0, 100, 255), 5)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 3)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
# first appearence of object with id=track.id
if track.track_id not in counter.people_init or counter.people_init[
track.track_id] == 0:
counter.obj_initialized(track.track_id)
rect_head = Rectangle(bbox[0], bbox[1], bbox[2], bbox[3])
intersection = rect_head & rect_door
if intersection:
intersection_square = rect_square(*intersection)
head_square = rect_square(*rect_head)
rat = intersection_square / head_square
# 1e-3 * frame.shape[0], (0, 100, 255), 5)
# was initialized in door, probably going in
if rat >= 0.7:
counter.people_init[track.track_id] = 2
# initialized in the bus, mb going out
elif rat <= 0.4 or bbox[3] > border_door:
counter.people_init[track.track_id] = 1
# initialized between the exit and bus, not obvious state
elif rat > 0.4 and rat < 0.7:
counter.people_init[track.track_id] = 3
counter.rat_init[track.track_id] = rat
# res is None, means that object is not in door contour
else:
counter.people_init[track.track_id] = 1
counter.people_bbox[track.track_id] = bbox
counter.cur_bbox[track.track_id] = bbox
adc = "%.2f" % (track.adc *
100) + "%" # Average detection confidence
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 5)
if not show_detections:
track_cls = track.cls
cv2.putText(frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.putText(
frame, 'ADC: ' + adc,
(int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])),
0, 1e-3 * frame.shape[0], (0, 255, 0), 1)
id_get_lost = [
track.track_id for track in tracker.tracks
if track.time_since_update >= 35
]
# and track.age >= 29]
# id_inside_tracked = [track.track_id for track in tracker.tracks if track.age > 60]
for val in counter.people_init.keys():
# check bbox also
inter_square = 0
cur_square = 0
ratio = 0
cur_c = find_centroid(counter.cur_bbox[val])
init_c = find_centroid(counter.people_bbox[val])
vector_person = (cur_c[0] - init_c[0], cur_c[1] - init_c[1])
if val in id_get_lost and counter.people_init[val] != -1:
rect_сur = Rectangle(counter.cur_bbox[val][0],
counter.cur_bbox[val][1],
counter.cur_bbox[val][2],
counter.cur_bbox[val][3])
inter = rect_сur & rect_door
if inter:
inter_square = rect_square(*inter)
cur_square = rect_square(*rect_сur)
try:
ratio = inter_square / cur_square
except ZeroDivisionError:
ratio = 0
# if vector_person < 0 then current coord is less than initialized, it means that man is going
# in the exit direction
if vector_person[1] > 70 and counter.people_init[val] == 2 \
and ratio < 0.9: # and counter.people_bbox[val][3] > border_door \
counter.get_in()
elif vector_person[1] < -70 and counter.people_init[val] == 1 \
and ratio >= 0.6:
counter.get_out()
elif vector_person[1] < -70 and counter.people_init[val] == 3 \
and ratio > counter.rat_init[val] and ratio >= 0.6:
counter.get_out()
elif vector_person[1] > 70 and counter.people_init[val] == 3 \
and ratio < counter.rat_init[val] and ratio < 0.6:
counter.get_in()
counter.people_init[val] = -1
del val
ins, outs = counter.show_counter()
cv2.rectangle(frame, (0, 0), (250, 50), (0, 0, 0), -1, 8)
cv2.putText(frame, "in: {}, out: {} ".format(ins, outs), (10, 35),
0, 1e-3 * frame.shape[0], (255, 255, 255), 3)
cv2.namedWindow('video', cv2.WINDOW_NORMAL)
cv2.resizeWindow('video', 1422, 800)
cv2.imshow('video', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %f" % fps)
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
del door_array
mean_error = np.mean(error_values)
print("mean error for {} videos: {}".format(len(files), mean_error))
def run(self):
asyncVideo_flag = self.args.asyncVideo_flag
writeVideo_flag = self.args.writeVideo_flag
if asyncVideo_flag:
video_capture = VideoCaptureAsync(self.video_path)
else:
video_capture = cv2.VideoCapture(self.video_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
count_frame = 0
while True:
count_frame += 1
t1 = time.time()
ret, frame = video_capture.read()
if ret != True:
break
_frame = frame
# process
frame = self.process(frame, _frame, count_frame)
# visualize
if self.args.visualize:
frame = imutils.resize(frame, width=1000)
cv2.imshow("Final result", frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 1.0
output_format = 'mp4'
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
show_detections = True
writeVideo_flag = True
asyncVideo_flag = False
fpeses = []
check_gpu()
video_name = 'test1.mp4'
print("opening video: {}".format(video_name))
file_path = join('data_files/videos', video_name)
output_name = 'save_data/out_' + video_name[0:-3] + output_format
counter = Counter(counter_in=0, counter_out=0, track_id=0)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
if writeVideo_flag:
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(output_name, fourcc, 15, (w, h))
left_array = [0, 0, w / 2, h]
fps = 0.0
fps_imutils = imutils.video.FPS().start()
rect_left = Rectangle(left_array[0], left_array[1], left_array[2],
left_array[3])
border_door = left_array[3]
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
with open('log_results.txt', 'a') as log:
log.write('1')
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls,
feature) for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
cv2.rectangle(frame, (int(left_array[0]), int(left_array[1])),
(int(left_array[2]), int(left_array[3])), (23, 158, 21),
3)
if len(detections) != 0:
counter.someone_inframe()
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
score = "%.2f" % (det.confidence * 100) + "%"
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 3)
else:
if counter.need_to_clear():
counter.clear_all()
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
if track.track_id not in counter.people_init or counter.people_init[
track.track_id] == 0:
counter.obj_initialized(track.track_id)
ratio_init = find_ratio_ofbboxes(bbox=bbox,
rect_compare=rect_left)
if ratio_init > 0:
if ratio_init >= 0.8 and bbox[3] < left_array[3]:
counter.people_init[
track.track_id] = 2 # man in left side
elif ratio_init < 0.8 and bbox[3] > left_array[
3]: # initialized in the bus, mb going out
counter.people_init[track.track_id] = 1
else:
counter.people_init[track.track_id] = 1
counter.people_bbox[track.track_id] = bbox
counter.cur_bbox[track.track_id] = bbox
adc = "%.2f" % (track.adc *
100) + "%" # Average detection confidence
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1]) + 50), 0,
1e-3 * frame.shape[0], (0, 255, 0), 5)
if not show_detections:
track_cls = track.cls
cv2.putText(frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.putText(
frame, 'ADC: ' + adc,
(int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
id_get_lost = [
track.track_id for track in tracker.tracks
if track.time_since_update >= 5
]
# TODO clear people_init and other dicts
for val in counter.people_init.keys():
ratio = 0
cur_c = find_centroid(counter.cur_bbox[val])
init_c = find_centroid(counter.people_bbox[val])
vector_person = (cur_c[0] - init_c[0], cur_c[1] - init_c[1])
if val in id_get_lost and counter.people_init[val] != -1:
ratio = find_ratio_ofbboxes(bbox=counter.cur_bbox[val],
rect_compare=rect_left)
if vector_person[0] > 200 and counter.people_init[val] == 2 \
and ratio < 0.7: # and counter.people_bbox[val][3] > border_door \
counter.get_out()
print(vector_person[0], counter.people_init[val], ratio)
elif vector_person[0] < -100 and counter.people_init[val] == 1 \
and ratio >= 0.7:
counter.get_in()
print(vector_person[0], counter.people_init[val], ratio)
counter.people_init[val] = -1
del val
ins, outs = counter.show_counter()
cv2.rectangle(frame, (700, 0), (950, 50), (0, 0, 0), -1, 8)
cv2.putText(frame, "in: {}, out: {} ".format(ins, outs), (710, 35), 0,
1e-3 * frame.shape[0], (255, 255, 255), 3)
cv2.namedWindow('video', cv2.WINDOW_NORMAL)
# cv2.resizeWindow('video', 1422, 800)
cv2.imshow('video', frame)
if writeVideo_flag:
out.write(frame)
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % fps)
if len(fpeses) < 15:
fpeses.append(round(fps, 2))
elif len(fpeses) == 15:
# fps = round(np.median(np.array(fpeses)))
median_fps = float(np.median(np.array(fpeses)))
fps = round(median_fps, 1)
print('max fps: ', fps)
fps = 20
counter.fps = fps
fpeses.append(fps)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
frame_wid = 1280
frame_hei = 720
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
# tracker = Tracker(metric)
show_detections = True
writeVideo_flag = True
asyncVideo_flag = False
# 存放所有视频数据的上级目录
#root_video_path = '/home/yhzn/ws/clip/'
root_video_path = '/home/yhzn/ws/citywalks/'
all_video_path = []
read_video(root_video_path, all_video_path)
break_flag = 0
result = []
myvideo_result = []
print("Video number = ", len(all_video_path))
just_video_cnt = 0
# 依次便利所有视频的路径并读取，each_video代表的是str类型路径
for each_video in all_video_path:
tracker = Tracker(metric)
file_path = each_video
# 将城市名与视频编号识别出来
video_name_split = file_path.split('/')
city_name = video_name_split[-2]
video_number = video_name_split[-1]
print(city_name, video_number)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
frame_num = 0
# 确认保存路径的存在
city_dir = ''.join(['/home/yhzn/ws/crop_image/', city_name])
if not os.path.exists(city_dir):
os.mkdir(city_dir)
video_dir = ''.join(
['/home/yhzn/ws/crop_image/', city_name, '/', video_number])
if not os.path.exists(video_dir):
os.mkdir(video_dir)
# 计数功能，仅仅为了显示处理进度用
just_frame_cnt = 0
just_video_cnt += 1
while True:
ret, frame = video_capture.read()
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# 检测框是否显示
# for det in detections:
# bbox = det.to_tlbr()
# if show_detections and len(classes) > 0:
# det_cls = det.cls
# score = "%.2f" % (det.confidence * 100) + "%"
# cv2.putText(frame, str(det_cls) + " " + score, (int(bbox[0]), int(bbox[3])), 0,
# 1e-3 * frame.shape[0], (0, 0, 255), 1)
# cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
# 仅仅使用了跟踪框
tmp_frame = frame.copy()
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
# print("ord1: ", bbox[0], bbox[1], bbox[2], bbox[3])
if bbox[0] < 0:
bbox[0] = 0
if bbox[0] >= frame_wid:
bbox[0] = frame_wid - 1
if bbox[1] < 0:
bbox[1] = 0
if bbox[1] >= frame_hei:
bbox[1] = frame_hei - 1
if bbox[2] < 0:
bbox[2] = 0
if bbox[2] >= frame_wid:
bbox[2] = frame_wid - 1
if bbox[3] < 0:
bbox[3] = 0
if bbox[3] >= frame_hei:
bbox[3] = frame_hei - 1
# print("ord2: ", bbox[0], bbox[1], bbox[2], bbox[3])
# if int(bbox[0]) < 0 or int(bbox[1]) < 0 or int(bbox[2]) < 0 or int(bbox[3]) < 0:
# continue
# 行人图像的裁剪
crop_image = frame[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])] # 裁剪
crop_image = cv2.resize(crop_image, (128, 256))
cv2.imwrite(
video_dir + '/frame_' + str(frame_num).zfill(4) + '_ped_' +
str(track.track_id) + '.png', crop_image)
# Average detection confidence
adc = "%.2f" % (track.adc * 100) + "%"
cv2.rectangle(tmp_frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (0, 255, 0), 2)
cv2.putText(tmp_frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1e-3 * frame.shape[0], (0, 0, 255), 1)
if not show_detections:
track_cls = track.cls
cv2.putText(tmp_frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.putText(
tmp_frame, 'ADC: ' + adc,
(int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])),
0, 1e-3 * frame.shape[0], (0, 255, 0), 1)
cx = int((int(bbox[0]) + int(bbox[2])) / 2)
cy = int((int(bbox[1]) + int(bbox[3])) / 2)
w = int(int(bbox[2]) - int(bbox[0]))
h = int(int(bbox[3]) - int(bbox[1]))
# 下面所有的append操作是为了保存所有跟踪到的结果，方便后需存放npy文件
tmp_result = []
tmp_result.append(video_number)
tmp_result.append(city_name)
tmp_result.append(str(frame_num))
tmp_result.append(str(track.track_id))
tmp_result.append(str(cx))
tmp_result.append(str(cy))
tmp_result.append(str(w))
tmp_result.append(str(h))
tmp_result.append('0')
tmp_result.append('0')
tmp_result.append('0')
if h > 50:
result.append(tmp_result)
# my video detection
# 这个文件是保存成DTP项目所需要的格式，与上面的保存大致类似，测试用
tmp_myvideo = []
tmp_myvideo.append(city_name + '/' + video_number)
tmp_myvideo.append(str(frame_num))
tmp_myvideo.append(int(bbox[0]))
tmp_myvideo.append(int(bbox[1]))
tmp_myvideo.append(int(bbox[2]))
tmp_myvideo.append(int(bbox[3]))
# track
tmp_myvideo.append(str(track.track_id))
# detection length
tmp_myvideo.append(str(0))
# Height
tmp_myvideo.append(str(h))
if h > 50:
myvideo_result.append(tmp_myvideo)
frame_num += 1
# cv2.imshow('', tmp_frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
# 添加一些提示信息
just_frame_cnt += 1
fps = (fps + (1. / (time.time() - t1))) / 2
print("[%d / %d]" % (just_video_cnt, len(all_video_path)), end=' ')
print("[frame: %d] [fps: %f]" %
(just_frame_cnt, (fps + (1. / (time.time() - t1))) / 2))
# 显示FPS
# if not asyncVideo_flag:
# fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %f" % (fps))
# Press Q to stop!
#if cv2.waitKey(1) & 0xFF == ord('q'):
# break_flag = 1
# break
#if break_flag == 1:
# break
'''
带label的跟踪结果保存
'''
f = open('./clip_data/tracking_result.csv', 'w')
csv_writer = csv.writer(f)
# 写入csv第一行
csv_writer.writerow([
"vid", "filename", "frame_num", "track", "cx", "cy", "w", "h",
"track_length", "labeled", "requires_features"
])
# 按照行人序号、城市名的顺序排序，帧数自动排序了
result.sort(key=lambda x: (x[1], eval(x[3])))
# 处理track条目
for i in range(len(result)):
if i == 0:
pass
else:
if result[i][3] == result[i - 1][3] and eval(
result[i][2]) == eval(result[i - 1][2]) + 1:
result[i][8] = str(eval(result[i - 1][8]) + 1)
else:
result[i][8] = '0'
# 处理labeled and require_feature条目
for i in range(len(result)):
# 不满足过去30帧、并且存在未来60帧的不进行预测，因为数据帧不足
if i <= 28 or i >= len(result) - 60:
pass
# csv_writer.writerow(result[i])
else:
track_index_now = eval(result[i][8])
track_index_pre29 = eval(result[i - 29][8])
track_index_post60 = eval(result[i + 60][8])
if result[i][3] == result[i - 29][3] and result[i][3] == result[i + 60][
3] and track_index_now == track_index_pre29 + 29 and track_index_now == track_index_post60 - 60:
result[i][9] = '1'
result[i][10] = '1'
for j in range(i - 29, i):
result[j][10] = '1'
# csv_writer.writerow(result[i])
for i in range(len(result)):
csv_writer.writerow(result[i])
# 写DTP的tracking result csv
train_file1 = open('./clip_data/myvideo_yolo_detection_train1.csv', 'w')
train_file2 = open('./clip_data/myvideo_yolo_detection_train2.csv', 'w')
train_file3 = open('./clip_data/myvideo_yolo_detection_train3.csv', 'w')
val_file1 = open('./clip_data/myvideo_yolo_detection_val1.csv', 'w')
val_file2 = open('./clip_data/myvideo_yolo_detection_val2.csv', 'w')
val_file3 = open('./clip_data/myvideo_yolo_detection_val3.csv', 'w')
test_file1 = open('./clip_data/myvideo_yolo_detection_test1.csv', 'w')
test_file2 = open('./clip_data/myvideo_yolo_detection_test2.csv', 'w')
test_file3 = open('./clip_data/myvideo_yolo_detection_test3.csv', 'w')
# 定义csv_writer
csv_train1 = csv.writer(train_file1)
csv_train1.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_train2 = csv.writer(train_file2)
csv_train2.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_train3 = csv.writer(train_file3)
csv_train3.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_val1 = csv.writer(val_file1)
csv_val1.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_val2 = csv.writer(val_file2)
csv_val2.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_val3 = csv.writer(val_file3)
csv_val3.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_test1 = csv.writer(test_file1)
csv_test1.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_test2 = csv.writer(test_file2)
csv_test2.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
csv_test3 = csv.writer(test_file3)
csv_test3.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
for i in range(len(myvideo_result)):
tmp_path = myvideo_result[i][0]
city = tmp_path.split('/')[0]
# train
if city in Train_cities1:
csv_train1.writerow(myvideo_result[i])
if city in Train_cities2:
csv_train2.writerow(myvideo_result[i])
if city in Train_cities3:
csv_train3.writerow(myvideo_result[i])
# val
if city in Validation_cities1:
csv_val1.writerow(myvideo_result[i])
if city in Validation_cities2:
csv_val2.writerow(myvideo_result[i])
if city in Validation_cities3:
csv_val3.writerow(myvideo_result[i])
# test
if city in Test_cities1:
csv_test1.writerow(myvideo_result[i])
if city in Test_cities2:
csv_test2.writerow(myvideo_result[i])
if city in Test_cities3:
csv_test3.writerow(myvideo_result[i])
f.close()
train_file1.close()
train_file2.close()
train_file3.close()
val_file1.close()
val_file2.close()
val_file3.close()
test_file1.close()
test_file2.close()
test_file3.close()
'''
写入STED模型所需要的.npy文件
size = (n, 8, 25)
'''
# box numpy数据保存
train_val_test_box(result)
# output label数据保存
train_val_test_label(result)
"""
my video detection write
保存成DTP需要的跟踪格式
"""
f = open('./clip_data/myvideo_yolo_detection.csv', 'w')
csv_writer = csv.writer(f)
csv_writer.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
# 按照跟踪序号排序
myvideo_result.sort(key=lambda x: x[6])
# 处理track条目
for i in range(len(myvideo_result)):
if i == 0:
pass
else:
# 如何filename相同，track序号相同，帧数连续则表示跟踪
if myvideo_result[i][6] == myvideo_result[i - 1][6] and eval(
myvideo_result[i][1]) == eval(
myvideo_result[i - 1][1]
) + 1 and myvideo_result[i][0] == myvideo_result[i - 1][0]:
myvideo_result[i][7] = str(eval(myvideo_result[i - 1][7]) + 1)
else:
myvideo_result[i][7] = '0'
for i in range(len(myvideo_result)):
csv_writer.writerow(myvideo_result[i])
f.close()
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename,batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
asyncVideo_flag = False
file_path = 'video.webm'
if asyncVideo_flag :
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[...,::-1]) # bgr to rgb
boxs = yolo.detect_image(image)[0]
confidence = yolo.detect_image(image)[1]
features = encoder(frame,boxs)
detections = [Detection(bbox, confidence, feature) for bbox, confidence, feature in zip(boxs, confidence, features)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,255,255), 2)
cv2.putText(frame, str(track.track_id),(int(bbox[0]), int(bbox[1])),0, 5e-3 * 200, (0,255,0),2)
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2)
cv2.rectangle(frame,(int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])),(255,0,0), 2)
cv2.putText(frame, score + '%', (int(bbox[0]), int(bbox[3])), 0, 5e-3 * 130, (0,255,0),2)
cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
fps = (fps + (1./(time.time()-t1))) / 2
print("FPS = %f"%(fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = False
asyncVideo_flag = False
webcamera_flag = True
ipcamera_flag = False
udp_flag = True
file_path = '/workspace/data/C0133_v4.mp4'
if asyncVideo_flag :
video_capture = VideoCaptureAsync(file_path)
elif ipcamera_flag :
video_capture = cv2.VideoCapture('rtsp://*****:*****@192.168.2.201/ONVIF/MediaInput?profile=def_profile1')
elif webcamera_flag :
video_capture = cv2.VideoCapture(0)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
if udp_flag:
HOST = ''
PORT = 5000
address = '192.168.2.255'
sock =socket(AF_INET, SOCK_DGRAM)
sock.setsockopt(SOL_SOCKET, SO_BROADCAST, 1)
sock.bind((HOST, PORT))
fps = 0.0
fps_imutils = imutils.video.FPS().start()
savetime = 0
while True:
nowtime = datetime.datetime.now().isoformat()
ret, frame = video_capture.read() # frame shape 640*480*3
t1 = time.time()
if time.time() - savetime >= 30:
print('save data')
cv2.imwrite("/workspace/images/image.png", frame)
savetime = time.time()
image = Image.fromarray(frame[...,::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [Detection(bbox, confidence, cls, feature) for bbox, confidence, cls, feature in
zip(boxes, confidence, classes, features)]
else:
detections = [Detection_YOLO(bbox, confidence, cls) for bbox, confidence, cls in
zip(boxes, confidence, classes)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id), (int(bbox[0]), int(bbox[1])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
# socket
message = str(nowtime + "," + str(track.track_id) + "," + str(int(bbox[0])) + "," + str(int(bbox[1])) + "," + str(int(bbox[2])) + "," + str(int(bbox[3])))
bmessage = message.encode('utf-8')
print(type(bmessage))
if udp_flag:
sock.sendto(message.encode('utf-8'), (address, PORT))
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2) + "%"
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
if len(classes) > 0:
cls = det.cls
cv2.putText(frame, str(cls) + " " + score, (int(bbox[0]), int(bbox[3])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
#cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1./(time.time()-t1))) / 2
print("FPS = %f"%(fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.2
nn_budget = None
nms_max_overlap = 1.0
output_format = 'mp4'
video_name = 'bus4_2in_4out.mp4'
file_path = join('data_files/videos', video_name)
output_name = 'save_data/out_' + video_name[0:-3] + output_format
initialize_door_by_yourself = False
door_array = None
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
show_detections = True
writeVideo_flag = True
asyncVideo_flag = False
counter = Counter(counter_in=0, counter_out=0, track_id=0)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(output_name, fourcc, 15, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
ret, first_frame = video_capture.read()
if door_array is None:
if initialize_door_by_yourself:
door_array = select_object(first_frame)[0]
print(door_array)
else:
all_doors = read_door_info('data_files/doors_info.csv')
door_array = all_doors[video_name]
border_door = door_array[3]
error_values = []
truth = get_truth(video_name)
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
total_count = counter.return_total_count()
true_total = truth.inside + truth.outside
err = abs(total_count - true_total) / true_total
log_res = "in video: {}\n predicted / true\n counter in: {} / {}\n counter out: {} / {}\n" \
" total: {} / {}\n error: {}\n______________\n".format(video_name, counter.counter_in,
truth.inside,
counter.counter_out, truth.outside,
total_count, true_total, err)
with open('log_results.txt', 'w') as file:
file.write(log_res)
print(log_res)
error_values.append(err)
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls,
feature) for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
cv2.rectangle(frame, (int(door_array[0]), int(door_array[1])),
(int(door_array[2]), int(door_array[3])), (23, 158, 21),
2)
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
score = "%.2f" % (det.confidence * 100) + "%"
rect_head = Rectangle(bbox[0], bbox[1], bbox[2], bbox[3])
rect_door = Rectangle(int(door_array[0]), int(door_array[1]),
int(door_array[2]), int(door_array[3]))
intersection = rect_head & rect_door
if intersection:
squares_coeff = rect_square(*intersection) / rect_square(
*rect_head)
cv2.putText(
frame,
score + " inter: " + str(round(squares_coeff, 3)),
(int(bbox[0]), int(bbox[3])), 0, 1e-3 * frame.shape[0],
(0, 100, 255), 5)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 3)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
# first appearence of object with id=track.id
if track.track_id not in counter.people_init or counter.people_init[
track.track_id] == 0:
counter.obj_initialized(track.track_id)
rect_head = Rectangle(bbox[0], bbox[1], bbox[2], bbox[3])
rect_door = Rectangle(door_array[0], door_array[1],
door_array[2], door_array[3])
res = rect_head & rect_door
if res:
inter_square = rect_square(*res)
head_square = rect_square(*rect_head)
# was initialized in door, probably going in
if (inter_square / head_square) >= 0.8:
counter.people_init[track.track_id] = 2
# initialized in the bus, mb going out
elif (inter_square /
head_square) <= 0.4 or bbox[3] > border_door:
counter.people_init[track.track_id] = 1
# res is None, means that object is not in door contour
else:
counter.people_init[track.track_id] = 1
counter.people_bbox[track.track_id] = bbox
counter.cur_bbox[track.track_id] = bbox
adc = "%.2f" % (track.adc *
100) + "%" # Average detection confidence
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 1e-3 * frame.shape[0],
(0, 255, 0), 5)
if not show_detections:
track_cls = track.cls
cv2.putText(frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.putText(
frame, 'ADC: ' + adc,
(int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
id_get_lost = [
track.track_id for track in tracker.tracks
if track.time_since_update >= 25 and track.age >= 29
]
id_inside_tracked = [
track.track_id for track in tracker.tracks if track.age > 60
]
for val in counter.people_init.keys():
# check bbox also
cur_c = find_centroid(counter.cur_bbox[val])
init_c = find_centroid(counter.people_bbox[val])
vector_person = (cur_c[0] - init_c[0], cur_c[1] - init_c[1])
if val in id_get_lost and counter.people_init[val] != -1:
# if vector_person < 0 then current coord is less than initialized, it means that man is going
# in the exit direction
if vector_person[1] > 70 and counter.people_init[
val] == 2: # and counter.people_bbox[val][3] > border_door \
counter.get_in()
elif vector_person[1] < -70 and counter.people_init[val] == 1:
counter.get_out()
counter.people_init[val] = -1
print(f"person left frame")
print(f"current centroid - init : {cur_c} - {init_c}\n")
print(f"vector: {vector_person}\n")
del val
# elif val in id_inside_tracked and val not in id_get_lost and counter.people_init[val] == 1 \
# and bb_intersection_over_union(counter.cur_bbox[val], door_array) <= 0.3 \
# and vector_person[1] > 0: # and \
# # counter.people_bbox[val][3] > border_door:
# counter.get_in()
#
# counter.people_init[val] = -1
# print(f"person is tracked for a long time")
# print(f"current centroid - init : {cur_c} - {init_c}\n")
# print(f"vector: {vector_person}\n")
# imaggg = cv2.line(frame, find_centroid(counter.cur_bbox[val]),
# find_centroid(counter.people_bbox[val]),
# (0, 0, 255), 7)
# cv2.imshow('frame', imaggg)
# cv2.waitKey(0)
ins, outs = counter.show_counter()
cv2.putText(frame, "in: {}, out: {} ".format(ins, outs), (10, 30), 0,
1e-3 * frame.shape[0], (255, 0, 0), 5)
cv2.namedWindow('image', cv2.WINDOW_NORMAL)
cv2.resizeWindow('image', 1400, 800)
cv2.imshow('image', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
mean_error = np.mean(error_values)
print("mean error for {} video: {}".format(video_name, mean_error))
def main():
parser = argparse.ArgumentParser(description='Process some integers.')
parser.add_argument('input', type=str, help='Input video path')
parser.add_argument('bbox', type=str, help='Input bounding box path')
parser.add_argument('output', type=str, help='Ouput video path')
args = parser.parse_args()
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = True
asyncVideo_flag = False
file_path = args.input
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(args.output, fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
# fps_imutils = imutils.video.FPS().start()
with open(args.bbox) as f:
data = json.load(f)
frame_index = 0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes = np.asarray(
[pred['bbox'] for pred in data[frame_index]['annotations']])
confidence = np.asarray(
[pred['score'] for pred in data[frame_index]['annotations']])
classes = np.asarray(
[pred['label'] for pred in data[frame_index]['annotations']])
if tracking:
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2) + "%"
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
if len(classes) > 0:
cls = det.cls
cv2.putText(frame,
str(cls) + " " + score,
(int(bbox[0]), int(bbox[3])), 0,
1.5e-3 * frame.shape[0], (0, 255, 0), 1)
# cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
# fps_imutils.update()
# if not asyncVideo_flag:
# fps = (fps + (1./(time.time()-t1))) / 2
# print("FPS = %f"%(fps))
# Press Q to stop!
# if cv2.waitKey(1) & 0xFF == ord('q'):
# break
# fps_imutils.stop()
# print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.25
max_cross_cosine_distance = 0.5
nn_budget = None
nms_max_overlap = 0.5
frame_rate = 12
file_path = 'reid-wide2'
file_path2 = 'reid-long2'
show_detections = False
writeVideo_flag = False
asyncVideo_flag = False
beta_calulate_flag = False
predict_ns_flag = False
predict_time = 0.5
multi_camera_flag = True
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric, frame_rate=frame_rate)
if multi_camera_flag:
metric2 = nn_matching.NearestNeighborDistanceMetric(
"cosine", max_cosine_distance, nn_budget)
tracker2 = Tracker(metric2, frame_rate=frame_rate)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path + ".mp4")
else:
video_capture = cv2.VideoCapture(file_path + ".mp4")
if multi_camera_flag:
video_capture2 = cv2.VideoCapture(file_path2 + ".mp4")
if asyncVideo_flag:
video_capture.start()
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
if writeVideo_flag:
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(file_path2 + "-output.avi", fourcc, frame_rate,
(w, h))
frame_index = -1
if multi_camera_flag:
out2 = cv2.VideoWriter(file_path + "-output2.avi", fourcc,
frame_rate, (w, h))
fps = 0.0
fps_imutils = imutils.video.FPS().start()
alpha = np.arctan((326 + (369 - 326) * (250 - 110) /
(300 - 110) - w / 2) / w * 6.4 / 3.6) * 180 / np.pi
beta = 0
beta_last = 0
x_temp = 0
track_angle = np.array([])
x_axis = np.array([])
x_predict = np.array([])
x_current = np.array([])
x_kalman = np.array([])
matches_id = []
alert_mode_flag = False
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls,
feature) for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
if predict_ns_flag:
tracker.predict_ns(predict_time)
if alert_mode_flag:
if len(tracker.tracks) > p:
ret = tracker.tracks[track1_idx].is_confirmed()
if not ret:
alert_mode_flag = False
else:
alert_mode_flag = False
if not alert_mode_flag:
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
det_cls = det.cls
score = "%.2f" % (det.confidence * 100) + "%"
cv2.putText(frame,
str(det_cls) + " " + score,
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 5:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
if predict_ns_flag:
if track.x_predict > 0 and track.x_predict < w:
cv2.circle(frame, (track.x_predict, int(h / 2)), 15,
(0, 0, 255), -1)
x_temp += 1 / frame_rate
x_axis = np.append(x_axis, x_temp)
x_predict = np.append(x_predict, track.x_predict)
x_current = np.append(x_current, track.x_current)
x_kalman = np.append(x_kalman, track.mean[0])
if beta_calulate_flag:
beta = np.arctan(
(track.mean[0] - w / 2) / w * 6.4 / 16) * 180 / np.pi
else:
bbox = tracker.tracks[track1_idx].to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (0, 0, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 1e-3 * frame.shape[0],
(0, 255, 0), 1)
# if not show_detections:
# track_cls = track.cls
# adc = "%.2f" % (track.adc * 100) + "%" # Average detection confidence
# cv2.putText(frame, str(track_cls), (int(bbox[0]), int(bbox[3])), 0, 1e-3 * frame.shape[0], (0, 255, 0),
# 1)
# cv2.putText(frame, 'ADC: ' + adc, (int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])), 0,
# 1e-3 * frame.shape[0], (0, 255, 0), 1)
if beta_calulate_flag:
if np.isclose(beta, beta_last) and abs(beta) > 7:
beta = np.sign(beta) * np.arctan(3.2 / 16) * 180 / np.pi
cv2.putText(frame, "Beta_angle: " + '{:4.2f}'.format(beta),
(20, 20), 0, 1e-3 * frame.shape[0], (0, 0, 255), 1)
beta_last = beta
x_temp += 1 / frame_rate
x_axis = np.append(x_axis, x_temp)
track_angle = np.append(track_angle, alpha + beta)
cv2.imshow('camera1', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
if multi_camera_flag:
ret, frame = video_capture2.read() # frame shape 640*480*3
if ret != True:
break
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker2.predict()
tracker2.update(detections)
matches, unmatched_tracks_1, unmatched_tracks_2 = cross_tracker_match(
tracker, tracker2, max_cross_cosine_distance)
matches_id.clear()
for track_1_idx, track_2_idx in matches:
matches_id.append((tracker.tracks[track_1_idx].track_id,
tracker2.tracks[track_2_idx].track_id))
print("Matches:", matches_id)
if alert_mode_flag:
if len(tracker2.tracks) > track2_idx:
ret = tracker2.tracks[track2_idx].is_confirmed()
if not ret:
alert_mode_flag = False
else:
alert_mode_flag = False
if not alert_mode_flag:
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
det_cls = det.cls
score = "%.2f" % (det.confidence * 100) + "%"
cv2.putText(frame,
str(det_cls) + " " + score,
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])),
(255, 0, 0), 2)
for track in tracker2.tracks:
if not track.is_confirmed() or track.time_since_update > 5:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])),
(255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
if cv2.waitKey(1) & 0xFF == ord('t'):
roi = cv2.selectROI("Select Target", frame)
Roi = [Detection(roi, 0, None, 0)]
ret, track2_idx = ROI_target_match(tracker2, Roi)
cv2.destroyWindow("Select Target")
if ret:
t_list = [i for (i, j) in matches if j == track2_idx]
if t_list != []:
track1_idx = t_list[0]
alert_mode_flag = True
else:
bbox = tracker2.tracks[track2_idx].to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (0, 0, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.imshow('camera2', frame)
if writeVideo_flag:
# save a frame
out2.write(frame)
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# if not asyncVideo_flag:
# fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %.2f" % (fps))
fps_imutils.update()
fps_imutils.stop()
print('imutils FPS: {:.2f}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
if multi_camera_flag:
out2.release()
if predict_ns_flag:
np.save("x_axis.npy", x_axis)
np.save("x_predict.npy", x_predict)
np.save("x_current.npy", x_current)
np.save("x_kalman.npy", x_kalman)
plt.plot(x_axis, x_current, label='X_Measured')
plt.plot(x_axis, x_kalman, label='X_Filtered')
plt.plot(x_axis, x_predict, label='X_Predicted')
plt.legend()
plt.xlabel('Time (s)')
plt.ylabel('Pixel')
plt.title('Motion Prediction')
plt.show()
if beta_calulate_flag:
np.save("x_axis.npy", x_axis)
np.save("track_angle.npy", track_angle)
plt.plot(x_axis, track_angle)
plt.xlabel('Time (s)')
plt.ylabel('Angle (deg)')
plt.title('Target Angle')
plt.show()
cv2.destroyAllWindows()
file_path = '/home/sohaibrabbani/PycharmProjects/Strong_Baseline_of_Pedestrian_Attribute_Recognition/videos/abc.mp4'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_custom.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
# L
# for img_path in tqdm(img_paths):
count = 0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
asyncVideo_flag = False
showImg_flag = True
file_path = 'C:\\Users\\Divided\\Desktop\\test_vid.mp4'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
while True:
preprocess_time_start = get_current_ms()
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
overlay = frame.copy()
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
preprocess_time = get_current_ms() - preprocess_time_start
# print("Preprocessing time:\t{}".format(preprocess_time))
detection_time_start = get_current_ms()
boxs = yolo.detect_image(image)[0]
confidence = yolo.detect_image(image)[1]
detection_time = get_current_ms() - detection_time_start
# print("Detection time:\t{}".format(detection_time))
features = encoder(frame, boxs)
detections = [Detection(bbox, confidence, feature) for bbox, confidence, feature in
zip(boxs, confidence, features)]
nms_start_time = get_current_ms()
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
nms_time = get_current_ms() - nms_start_time
# print("NMS time:\t{}".format(nms_time))
tracker_start_time = get_current_ms()
# Call the tracker
tracker.predict()
tracker.update(detections)
tracker_time = get_current_ms() - tracker_start_time
# print("Tracker time:\t{}".format(tracker_time))
postprocess_start_time = get_current_ms()
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
track_bbox = track.to_tlbr()
cv2.rectangle(frame, (int(track_bbox[0]), int(track_bbox[1])),
(int(track_bbox[2]), int(track_bbox[3])), (255, 255, 255), 1)
if track.det_hist:
det_curr = track.det_hist[-1]
det_bbox = det_curr.to_tlbr()
score = "%.2f" % round(det_curr.confidence, 2)
cv2.rectangle(overlay, (int(det_bbox[0]), int(det_bbox[1])), (int(det_bbox[2]), int(det_bbox[3])),
track.color, 2)
cv2.putText(overlay, score, (int(det_bbox[0]) + 5, int(det_bbox[3]) - 5), 0, 0.5, (255, 255, 255),
2)
cv2.putText(overlay, str(track.track_id), (int(det_bbox[0]) + 5, int(det_bbox[1]) - 5), 0, 0.5,
(255, 255, 255), 2)
# centroid = det_curr.get_centroid()
# cv2.circle(overlay, (int(centroid[0]), int(centroid[1])), radius=3,
# color=(0, 0, 255), thickness=-1)
alpha = 0.5 # Transparency factor.
# Following line overlays transparent rectangle over the image
frame = cv2.addWeighted(overlay, alpha, frame, 1 - alpha, 0)
if showImg_flag:
cv2.imshow('', frame)
postprocess_time = get_current_ms() - postprocess_start_time
# print("Postprocess time:\t{}".format(postprocess_time))
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
fps = (fps + (1. / (time.time() - t1))) / 2
print("{}/{}/{}/{}/{}".format(preprocess_time, detection_time, nms_time, tracker_time, postprocess_time))
print("FPS = %f" % fps)
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
h = 0
w = 0
frame_index = -1
fps = 0.0
flows = OrderedDict()
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
writeVideo_flag = True
asyncVideo_flag = False
file_path = '/home/sohaibrabbani/Downloads/overhead_people_clips/baggageclaim.mp4'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov3.avi', fourcc,
video_capture.get(cv2.CAP_PROP_FPS), (w, h))
# frame_index = -1
fps_imutils = imutils.video.FPS().start()
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxs = yolo.detect_image(image)[0]
confidence = yolo.detect_image(image)[1]
features = encoder(frame, boxs)
detections = [
Detection(bbox, confidence, feature)
for bbox, confidence, feature in zip(boxs, confidence, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 60:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200,
(0, 255, 0), 2)
# Initializing the flow list for saving the flow data of each person - Sohaib
if track.track_id not in flows:
flows[track.track_id] = OrderedDict()
# Saving location of a person in a frame - Sohaib
flows[track.track_id][frame_index + 1] = np.array(
[int(bbox[0]), int(bbox[1])])
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
cv2.putText(frame, score + '%', (int(bbox[0]), int(bbox[3])), 0,
5e-3 * 130, (0, 255, 0), 2)
cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# Calling the serialize function to save the pickle file - Sohaib
serialize_flow_data(flows=flows,
height=h,
width=w,
file_path=file_path,
frame_count=video_capture.get(
cv2.CAP_PROP_FRAME_COUNT),
fps=video_capture.get(cv2.CAP_PROP_FPS))
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
show_detections = True
writeVideo_flag = True
asyncVideo_flag = False
file_path = '/media/wangsen/新加卷/MOF_data/citywalks/clips/PADUA/clip_000001.mp4'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
frame_num = 0
result = []
ped_cnt = 0
ped_index_store = []
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls,
feature) for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
# for det in detections:
# bbox = det.to_tlbr()
# if show_detections and len(classes) > 0:
# det_cls = det.cls
# score = "%.2f" % (det.confidence * 100) + "%"
# cv2.putText(frame, str(det_cls) + " " + score, (int(bbox[0]), int(bbox[3])), 0,
# 1e-3 * frame.shape[0], (0, 255, 0), 1)
# cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
if int(bbox[0]) < 0 or int(bbox[1]) < 0 or int(bbox[2]) < 0 or int(
bbox[3]) < 0:
continue
# 行人图像的裁剪
crop_image = frame[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])] # 裁剪
crop_image = cv2.resize(crop_image, (128, 256))
cv2.imwrite(
'./crop_image/frame_' + str(frame_num).zfill(4) + '_ped_' +
str(track.track_id) + '.png', crop_image)
adc = "%.2f" % (track.adc *
100) + "%" # Average detection confidence
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 1e-3 * frame.shape[0],
(0, 255, 0), 1)
if not show_detections:
track_cls = track.cls
cv2.putText(frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.putText(
frame, 'ADC: ' + adc,
(int(bbox[0]), int(bbox[3] + 2e-2 * frame.shape[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
w = int(int(bbox[2]) - int(bbox[0]))
h = int(int(bbox[3]) - int(bbox[1]))
tmp_result = []
# filename
tmp_result.append("clip_000001.mp4")
# frame_num
tmp_result.append(str(frame_num))
# bb1~bb4
tmp_result.append(int(bbox[0]))
tmp_result.append(int(bbox[1]))
tmp_result.append(int(bbox[2]))
tmp_result.append(int(bbox[3]))
# track
tmp_result.append(str(track.track_id))
# detection length
tmp_result.append(str(0))
# Height
tmp_result.append(str(h))
result.append(tmp_result)
frame_num += 1
cv2.imshow('', frame)
if writeVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# 正式写入
f = open('myvideo_yolo_detection.csv', 'w')
csv_writer = csv.writer(f)
# 写入头
csv_writer.writerow([
"filename", "frame_num", "bb1", "bb2", "bb3", "bb4", "track",
"detection_length", "Height"
])
# 按照跟踪序号排序
result.sort(key=lambda x: x[6])
# Item handle: track
for i in range(len(result)):
if i == 0:
pass
else:
# 如何filename相同，track序号相同，帧数连续则表示跟踪
if result[i][6] == result[i - 1][6] and eval(result[i][1]) == eval(result[i - 1][1]) + 1 and result[i][0] == \
result[i - 1][0]:
result[i][7] = str(eval(result[i - 1][7]) + 1)
else:
result[i][7] = '0'
for i in range(len(result)):
csv_writer.writerow(result[i])
f.close()
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = True
asyncVideo_flag = False
file_path = 'video.webm'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
model_par, valid_transform = model_init_par()
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
else:
detections = [
Detection_YOLO(bbox, confidence, cls)
for bbox, confidence, cls in zip(boxes, confidence, classes)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2) + "%"
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
if len(classes) > 0:
cls = det.cls
cv2.putText(frame,
str(cls) + " " + score,
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
#crop_img = frame[int(bbox[1]):int(bbox[3]),int(bbox[0]):int(bbox[2])]
crop_img = image.crop(
[int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3])])
#res_txt = demo_par(model_par, valid_transform, crop_img)
#draw.rectangle(xy=person_bbox[:-1], outline='red', width=1)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 1)
font = ImageFont.truetype(
'/home/sohaibrabbani/PycharmProjects/Strong_Baseline_of_Pedestrian_Attribute_Recognition/arial.ttf',
size=10)
# positive_cnt = 1
# for txt in res_txt:
# if 'personal' in txt:
# #draw.text((x1, y1 + 20 * positive_cnt), txt, (255, 0, 0), font=font)
# cv2.putText(frame, txt, (int(bbox[0]), int(bbox[1]) + 20 * positive_cnt), 0,
# 1e-3 * frame.shape[0], (0, 255, 0), 1)
# positive_cnt += 1
cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# 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)
tracker = Tracker(metric)
# Flags for process
tracking = True # Set False if you only want to do detection
writeVideo_flag = True # Set False if you don't want to write frames locally
asyncVideo_flag = False # It uses asynchronous processing for better FPS :Warning: Shuttering Problem
# Video File Path
file_path = '/mydrive/test.mp4'
# Check if asyncVideo flag set to True
if asyncVideo_flag :
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
output_file = os.path.basename(file_path)[:-4]
out = cv2.VideoWriter('./Output/' + output_file + "-prueba-test.avi", fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
while True:
ret, frame = video_capture.read() # Capture frames
if ret != True:
break
t1 = time.time()
# bgr to rgb frame conversion
image = Image.fromarray(frame[...,::-1])
# YOLOv4 Detection
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
# Encodes the frame and boxes for DeepSORT
features = encoder(frame, boxes)
# DeepSORT Detection
detections = [Detection(bbox, confidence, cls, feature) for bbox, confidence, cls, feature in
zip(boxes, confidence, classes, features)]
else:
# Only YOLOv4 Detection
detections = [Detection_YOLO(bbox, confidence, cls) for bbox, confidence, cls in
zip(boxes, confidence, classes)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
# Draw white bbox for DeepSORT
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id), (int(bbox[0]), int(bbox[1])), cv2.FONT_HERSHEY_COMPLEX_SMALL, 1.5,(0, 255, 0), 1)
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2)
# Check the class for colored bbox
if len(classes) > 0:
cls = det.cls
center_bbox = (int(bbox[2]), int(bbox[2]))
if str(cls) == 'person':
# Draw Blue bbox for YOLOv4 person detection
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (219, 152, 52), 2)
elif str(cls) == 'backpack' or 'handbag' or 'suitcase':
# Draw Orange bbox for YOLOv4 handbag, backpack and suitcase detection
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (65, 176, 245), 2)
if not asyncVideo_flag:
fps = (fps + (1./(time.time()-t1))) / 2
print("FPS = %f"%(fps))
cv2.putText(frame, "GPU: NVIDIA Tesla P100", (5, 70), cv2.FONT_HERSHEY_COMPLEX_SMALL, 0.8, (0, 255, 0), 1)
cv2.putText(frame, "FPS: %.2f" % fps, (5, 50), cv2.FONT_HERSHEY_COMPLEX_SMALL, 0.8, (0, 255, 0), 1)
# draw the timestamp on the frame
timestamp = datetime.datetime.now()
ts = timestamp.strftime("%d/%m/%Y, %H:%M:%S")
cv2.putText(frame, ts, (5, 30), cv2.FONT_HERSHEY_COMPLEX_SMALL, 0.8, (0, 255, 0), 1)
#cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
class CameraSource(object):
def __init__(self, cameraSource, height, output_file=None, startFrame=0,
async_read=False, outputToServer=False, capture_size=None):
if async_read:
self.camera = VideoCaptureAsync(cameraSource)
else:
self.camera = cv2.VideoCapture(cameraSource)
if capture_size is not None:
print(capture_size)
self.camera.set(cv2.CAP_PROP_FRAME_WIDTH, capture_size[0])
self.camera.set(cv2.CAP_PROP_FRAME_HEIGHT, capture_size[1])
if async_read:
self.ORIGINAL_WIDTH = self.camera.width
self.ORIGINAL_HEIGHT = self.camera.height
else:
self.ORIGINAL_WIDTH = int(self.camera.get(cv2.CAP_PROP_FRAME_WIDTH))
self.ORIGINAL_HEIGHT = int(self.camera.get(cv2.CAP_PROP_FRAME_HEIGHT))
print('CameraSource')
print('Requested capture size', capture_size)
print('Actual capture size', self.ORIGINAL_WIDTH, self.ORIGINAL_HEIGHT)
self.HEIGHT = height
self.WIDTH = self.ORIGINAL_WIDTH * self.HEIGHT // self.ORIGINAL_HEIGHT
self.WIDTH = self.WIDTH + self.WIDTH % 2 # Make it even.
self.startFrame = startFrame
self.nFrames = 0
self.writer = VideoWriter(output_file)
if async_read:
self.camera.start()
self.outputToServer = outputToServer
if outputToServer:
# https://robotpy.readthedocs.io/en/stable/vision/code.html
pass
#self.outputStream = CameraServer.getInstance().putVideo(
# 'ProcessedVisionFrame', self.WIDTH, self.HEIGHT)
def GetFrame(self):
# Processing on first call.
if self.nFrames == 0:
# Skip some frames if requested.
if self.startFrame > 0:
skippedFrames = 0
while True:
ret, frame = self.camera.read()
if not ret or frame is None:
print('No more frames')
return None
skippedFrames += 1
if skippedFrames >= self.startFrame:
break
# Start timer for first frame.
self.startTime = time.time()
# Get frame.
frame = None
frameTime = time.time()
if self.nFrames > 0 and self.nFrames % 50 == 0:
print('FPS: ', self.nFrames / (frameTime - self.startTime))
self.nFrames += 1
ret, frame = self.camera.read()
if ret and frame is not None:
if frame.shape[0] != self.HEIGHT:
frame = cv2.resize(frame, (self.WIDTH, self.HEIGHT))
return frame
def ImageSize(self):
return (self.WIDTH, self.HEIGHT)
def OutputFrameAndTestContinue(self, message, frame, height=None):
self.writer.OutputFrame(frame)
if self.outputToServer:
self.outputStream.putFrame(frame)
return ShowFrameAndTestContinue(message, frame, height)
def __del__(self):
self.camera.release()
cv2.destroyAllWindows()
def main():
PATH_TO_CKPT_PERSON = 'models/faster_rcnn_restnet50.pb'
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = 200
nms_max_overlap = 1.0
yolo = YOLO()
reid = PERSON_REID()
frozen_person = FROZEN_GRAPH_INFERENCE(PATH_TO_CKPT_PERSON)
# # Deep SORT
# model_filename = 'model_data/mars-small128.pb'
# encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
# file_path = 0
if VIDEO_CAPTURE == 0 and asyncVideo_flag == True:
video_capture = VideoCaptureAsync(file_path)
elif VIDEO_CAPTURE == 1 and asyncVideo_flag == True:
video_capture = myVideoCapture(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
im_width = int(video_capture.get(3))
im_height = int(video_capture.get(4))
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(output_filename, fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
boxs = list()
confidence = list()
persons = list()
frame_count = 0
track_count = 0
num_files = 0
while True:
t1 = time.time()
ret, frame = video_capture.read() # frame shape 640*480*3
frame_org = frame.copy()
if ret != True:
break
frame_count += 1
# print('Frame count: {}'.format(frame_count))
# Person detection using Frozen Graph
persons = frozen_person.run_frozen_graph(frame, im_width, im_height)
boxs = [[
person['left'], person['top'], person['width'], person['height']
] for person in persons]
confidence = [person['confidence'] for person in persons]
cropped_persons = list(person['cropped'] for person in persons)
# # Person detection using YOLO - Keras-converted model
# image = Image.fromarray(frame[...,::-1]) # bgr to rgb
# boxs = yolo.detect_image(image)[0]
# confidence = yolo.detect_image(image)[1]
# cropped_persons = [np.array(frame[box[1]:box[1]+box[3], box[0]:box[0]+box[2]]) for box in boxs] #[x,y,w,h]
# features = encoder(frame, boxs)
if len(cropped_persons) > 0:
features = reid.extract_feature_imgTensor(cropped_persons)
# print(features.shape)
detections = [
Detection(bbox, confidence,
feature) for bbox, confidence, feature in zip(
boxs, confidence, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (0, 0, 255), 2)
cv2.putText(frame, str(track.track_id),
(int(bbox[0]), int(bbox[1])), 0, 5e-3 * 200,
(0, 255, 0), 2)
directory = os.path.join('output', str(track.track_id))
if not os.path.exists(directory):
os.makedirs(directory)
# file_count = len([name for name in os.listdir(directory+'/') if os.path.isfile(name)])
file_count = sum(
[len(files) for root, dirs, files in os.walk(directory)])
# print(file_count)
if file_count == 0:
cv2.imwrite(
directory + '/' + str(file_count + 1) + '.jpg',
frame_org[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])])
elif file_count > 0 and track_count % 10 == 0:
cv2.imwrite(
directory + '/' + str(file_count + 1) + '.jpg',
frame_org[int(bbox[1]):int(bbox[3]),
int(bbox[0]):int(bbox[2])])
track_count += 1
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
cv2.putText(frame, score + '%', (int(bbox[0]), int(bbox[3])),
0, 5e-3 * 130, (0, 255, 0), 2)
cv2.imshow('YOLO DeepSort', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %f"%(fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
# print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
width = 1280
height = 720
rfps = 10
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
tracking = args.tracking
writeVideo_flag = args.writeVideo_flag
asyncVideo_flag = args.asyncVideo_flag
webcamera_flag = args.webcamera_flag
ipcamera_flag = args.ipcamera_flag
udp_flag = args.udp_flag
full_cam_addr, key = sd.set_address(args.ipaddress, args.cam_ip,
args.cam_cmd, args.key)
cam_ip = full_cam_addr.replace(args.cam_cmd,
"").replace("rtsp://*****:*****@", "")
if args.jpegmode:
full_cam_addr = full_cam_addr.replace("rtsp", "http")
print(full_cam_addr)
print(key)
if asyncVideo_flag:
print("load videofile")
video_capture = VideoCaptureAsync(args.videofile)
elif ipcamera_flag or args.jpegmode:
print("load ipcamera")
video_capture = cv2.VideoCapture(full_cam_addr)
# width = video_capture.get(cv2.CAP_PROP_FRAME_WIDTH)
# height = video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT)
# rfps = video_capture.get(cv2.CAP_PROP_FPS)
print("fps:{}width:{}height:{}".format(rfps, width, height))
elif webcamera_flag:
print("load webcamera")
video_capture = cv2.VideoCapture(0)
else:
print("load videofile")
video_capture = cv2.VideoCapture(args.videofile)
# video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
if udp_flag:
HOST = ''
PORT = 5000
address = '192.168.2.255'
sock = socket(AF_INET, SOCK_DGRAM)
sock.setsockopt(SOL_SOCKET, SO_BROADCAST, 1)
sock.bind((HOST, PORT))
fps = 0.0
i = 0
if not args.maskoff:
maskbgi = Image.new('RGB', (int(width), int(height)), (0, 0, 0))
mask = Image.open(args.maskdir + 'mask' + args.ipaddress[-1] +
'.png').convert("L").resize(size=(int(width),
int(height)),
resample=Image.NEAREST)
while True:
nowtime = datetime.datetime.now(
timezone('Asia/Tokyo')).strftime('%Y-%m-%d %H:%M:%S.%f%z')
ret, frame = video_capture.read() # frame shape 640*480*3
if not ret:
print('cant read')
video_capture = cv2.VideoCapture(full_cam_addr)
continue
t1 = time.time()
try:
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
except TypeError:
video_capture = cv2.VideoCapture(full_cam_addr)
continue
image = Image.composite(maskbgi, image, mask)
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
else:
detections = [
Detection_YOLO(bbox, confidence, cls)
for bbox, confidence, cls in zip(boxes, confidence, classes)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
car_data = {}
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
car_data[str(track.track_id)] = [
int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3])
]
sd.send_amqp(
sd.create_jsondata(cam_ip, nowtime,
time.time() - t1, car_data, args.jsonfile,
args.json_path, i), key, args.AMQPHost)
i += 1
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
### 読み飛ばし処理を追加 ###
if not args.jsonfile and args.skip:
if fps <= 10:
for _i in range(int(math.ceil(rfps / fps)) - 1):
ret, frame = video_capture.read()
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
with open("cfg/detection_tracker_cfg.json") as detection_config:
detect_config = json.load(detection_config)
with open("cfg/doors_info.json") as doors_config:
doors_config = json.load(doors_config)
with open("cfg/around_doors_info.json") as around_doors_config:
around_doors_config = json.load(around_doors_config)
model_filename = detect_config["tracking_model"]
input_folder, output_folder = detect_config["input_folder"], detect_config[
"output_folder"]
meta_folder = detect_config["meta_folder"]
output_format = detect_config["output_format"]
# Deep SORT
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
show_detections = True
asyncVideo_flag = False
check_gpu()
# from here should start loop to process videos from folder
# for video_name in os.listdir(input_folder):
HOST = "localhost"
PORT = 8075
with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as sock:
sock.bind((HOST, PORT))
sock.listen()
conn, addr = sock.accept()
with conn:
print('Connected by', addr)
# loop over all videos
while True:
data = conn.recv(1000)
video_motion_list = data.decode("utf-8").split(';')
videos_que = deque()
for video_motion in video_motion_list:
videos_que.append(video_motion)
video_name = videos_que.popleft()
if not video_name.endswith(output_format):
continue
print('elements in que', len(videos_que))
print("opening video: {}".format(video_name))
full_video_path = join(input_folder, video_name)
# full_video_path = "rtsp://*****:*****@192.168.1.52:554/1/h264major"
meta_name = meta_folder + video_name[:-4] + ".json"
with open(meta_name) as meta_config_json:
meta_config = json.load(meta_config_json)
camera_id = meta_config["camera_id"]
if not os.path.exists(output_folder + str(camera_id)):
os.mkdir(output_folder + str(camera_id))
output_name = output_folder + camera_id + '/out_' + video_name
counter = Counter(counter_in=0, counter_out=0, track_id=0)
tracker = Tracker(metric)
if asyncVideo_flag:
video_capture = VideoCaptureAsync(full_video_path)
video_capture.start()
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
video_capture = cv2.VideoCapture(full_video_path)
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter(output_name, fourcc, 25, (w, h))
door_array = doors_config["{}".format(camera_id)]
around_door_array = tuple(
around_doors_config["{}".format(camera_id)])
rect_door = Rectangle(door_array[0], door_array[1],
door_array[2], door_array[3])
border_door = door_array[3]
# loop over video
save_video_flag = False
while True:
fps_imutils = imutils.video.FPS().start()
ret, frame = video_capture.read()
if not ret:
with open('videos_saved/log_results.txt', 'a') as log:
log.write(
'processed (ret). Time: {}, camera id: {}\n'.
format(video_name, camera_id))
break
t1 = time.time()
# lost_ids = counter.return_lost_ids()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
# image = image.crop(around_door_array)
boxes, confidence, classes = yolo.detect_image(image)
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
classes = np.array([d.cls for d in detections])
indices = preprocessing.non_max_suppression(
boxes, nms_max_overlap, scores)
detections = [detections[i] for i in indices]
# Call the tracker
tracker.predict()
tracker.update(detections)
cv2.rectangle(frame,
(int(door_array[0]), int(door_array[1])),
(int(door_array[2]), int(door_array[3])),
(23, 158, 21), 3)
if len(detections) != 0:
counter.someone_inframe()
for det in detections:
bbox = det.to_tlbr()
if show_detections and len(classes) > 0:
score = "%.2f" % (det.confidence * 100) + "%"
cv2.rectangle(frame,
(int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])),
(255, 0, 0), 3)
else:
if counter.need_to_clear():
counter.clear_all()
# identities = [track.track_id for track in tracker.tracks]
# counter.update_identities(identities)
for track in tracker.tracks:
if not track.is_confirmed(
) or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
if track.track_id not in counter.people_init or counter.people_init[
track.track_id] == 0:
# counter.obj_initialized(track.track_id)
ratio_init = find_ratio_ofbboxes(
bbox=bbox, rect_compare=rect_door)
if ratio_init > 0:
if ratio_init >= 0.5: # and bbox[3] < door_array[3]:
counter.people_init[
track.track_id] = 2 # man in the door
elif ratio_init < 0.5: # and bbox[3] > door_array[3]: # initialized in the outside
counter.people_init[track.track_id] = 1
else:
counter.people_init[track.track_id] = 1
counter.people_bbox[track.track_id] = bbox
counter.cur_bbox[track.track_id] = bbox
adc = "%.2f" % (track.adc * 100
) + "%" # Average detection confidence
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])),
(int(bbox[2]), int(bbox[3])),
(255, 255, 255), 2)
cv2.putText(frame, "ID: " + str(track.track_id),
(int(bbox[0]), int(bbox[1]) + 50), 0,
1e-3 * frame.shape[0], (0, 255, 0), 3)
if not show_detections:
track_cls = track.cls
cv2.putText(frame, str(track_cls),
(int(bbox[0]), int(bbox[3])), 0,
1e-3 * frame.shape[0], (0, 255, 0), 3)
cv2.putText(frame, 'ADC: ' + adc,
(int(bbox[0]),
int(bbox[3] + 2e-2 * frame.shape[1])),
0, 1e-3 * frame.shape[0], (0, 255, 0),
3)
# if track.time_since_update >= 15:
# id_get_lost.append(track.track_id)
id_get_lost = [
track.track_id for track in tracker.tracks
if track.time_since_update >= 15
]
for val in counter.people_init.keys():
ratio = 0
cur_c = find_centroid(counter.cur_bbox[val])
init_c = find_centroid(counter.people_bbox[val])
if val in id_get_lost and counter.people_init[
val] != -1:
ratio = find_ratio_ofbboxes(
bbox=counter.cur_bbox[val],
rect_compare=rect_door)
if counter.people_init[val] == 2 \
and ratio < 0.6: # and counter.people_bbox[val][3] > border_door \
counter.get_out()
save_video_flag = True
print(counter.people_init[val], ratio)
elif counter.people_init[val] == 1 \
and ratio >= 0.6:
counter.get_in()
save_video_flag = True
print(counter.people_init[val], ratio)
counter.people_init[val] = -1
ins, outs = counter.return_counter()
cv2.rectangle(frame, (frame.shape[1] - 150, 0),
(frame.shape[1], 50), (0, 0, 0), -1, 8)
cv2.putText(frame, "in: {}, out: {} ".format(ins, outs),
(frame.shape[1] - 140, 20), 0,
1e-3 * frame.shape[0], (255, 255, 255), 3)
out.write(frame)
fps_imutils.update()
if not asyncVideo_flag:
pass
# fps = (1. / (time.time() - t1))
# print("FPS = %f" % fps)
# if len(fpeses) < 15:
# fpeses.append(round(fps, 2))
#
# elif len(fpeses) == 15:
# # fps = round(np.median(np.array(fpeses)))
# median_fps = float(np.median(np.array(fpeses)))
# fps = round(median_fps, 1)
# print('max fps: ', fps)
# # fps = 20
# counter.fps = fps
# fpeses.append(fps)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
if asyncVideo_flag:
video_capture.stop()
del video_capture
else:
video_capture.release()
if save_video_flag:
with open('videos_saved/log_results.txt', 'a') as log:
log.write(
'detected!!! time: {}, camera id: {}, detected move in: {}, out: {}\n'
.format(video_name, camera_id, ins, outs))
log.write('video written {}\n\n'.format(output_name))
out.release()
else:
if out.isOpened():
out.release()
if os.path.isfile(output_name):
os.remove(output_name)
if os.path.isfile(full_video_path):
os.remove(full_video_path)
if os.path.isfile(meta_name):
os.remove(meta_name)
save_video_flag = False
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine", max_cosine_distance, nn_budget)
tracker = Tracker(metric)
np.random.seed(42)
COLORS = np.random.randint(0, 255, size=(200, 3),
dtype="uint8")
writeVideo_flag = True
asyncVideo_flag = False
file_path = 'testvideo2.avi'
# load the COCO class labels our YOLO model was trained on 加载我们的YOLO模型经过培训的COCO类标签
labelsPath = os.path.sep.join(["model_data", "coco_classes.txt"])
LABELS = open(labelsPath).read().strip().split("\n")
# print(str(len(LABELS))+"load successfully")
# print(LABELS)
class_nums = np.zeros(80)
counter = np.zeros(80)
track_id_max = -1
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('testvideo2_out.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
img_num = 0
frame_cnt = 0
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
frame_copy = frame.copy()
if ret != True or frame_cnt > 30:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxs, confidence, class_names = yolo.detect_image(image)
features = encoder(frame, boxs)
detections = [Detection(bbox, confidence, feature) for bbox, confidence, feature in
zip(boxs, confidence, features)]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap, scores)
# print("print indices!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
# print(indices)
# for i in indices:
# print(str(i)+class_names[i][0])
# print(indices.shape)
detections = [detections[i] for i in indices]
class_names = [class_names[i] for i in indices]
print("class_name:" + str(class_names))
class_IDs = []
current_nums = np.zeros(80)
# class_IDs=[]
for class_name in class_names:
for i, LABEL in enumerate(LABELS):
if class_name[0] == LABEL:
current_nums[i] += 1
class_IDs.append(i)
# print("person:"+str(current_nums[0]))
cv2.putText(frame, 'Current', (20, 70), cv2.FONT_HERSHEY_DUPLEX, 0.75, (255, 255, 255), 2)
cv2.putText(frame, 'Total', (180, 70), cv2.FONT_HERSHEY_DUPLEX, 0.75, (0, 255, 255), 2)
x1 = 20
y1 = 100
for i, cl in enumerate(current_nums):
if cl > 0:
cv2.putText(frame, LABELS[i] + "=" + str(cl), (x1, y1), cv2.FONT_HERSHEY_DUPLEX, 0.6, (255, 255, 255),
1) # 当前帧各类别数量
y1 = y1 + 20
for i, det in enumerate(detections):
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2)
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 1)
cv2.putText(frame, score + '%', (int(bbox[0]), int(bbox[1]) + 10), cv2.FONT_HERSHEY_DUPLEX, 0.3,
(255, 255, 255), 1)
# cv2.putText(frame, class_names[i],(int(bbox[0]), int(bbox[1])-5), 0, 5e-3 * 130, (0, 255, 0), 2)
# cv2.putText(frame, class_names[i], (int(bbox[0]), int(bbox[1])), 0, 5e-3 * 130,(0, 255, 255), 2)
print("Total of detections:" + str(len(detections)))
# Call the tracker
tracker.predict()
tracker.update(detections, class_IDs)
# for i, cl in enumerate(class_nums):
# if cl > 0:
# print("add: " + LABELS[i] + str(cl - class_last_nums[i]))
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
print("not track.is_confirmed() or track.time_since_update > 1: " + str(track.track_id))
continue
bbox = track.to_tlbr()
color = [int(c) for c in COLORS[track.class_id % len(COLORS)]]
cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), color, 2)
cv2.putText(frame, str(track.track_id) + ' ' + LABELS[track.class_id], (int(bbox[0]), int(bbox[1]) - 5),
cv2.FONT_HERSHEY_DUPLEX, 0.4, color, 1)
if track.track_id > track_id_max:
counter[track.class_id] = counter[track.class_id] + 1
track_id_max = track.track_id
dest = frame_copy[int(bbox[1]):int(bbox[3]), int(bbox[0]):int(bbox[2])]
pdest = "./output_image/" + str(LABELS[track.class_id]) + str(int(counter[track.class_id])) + ".png"
cv2.imwrite(pdest, dest)
img_num += 1
# class_nums[track.class_id].append(track.track_id)
# print(str(LABELS[track.class_id])+":"+class_nums[track.class_id])
# print("track.id: " + str(track.track_id))
# print("track.class_name: " + str(LABELS[track.class_id]))
print(str(counter))
print("--------------------------该帧输出完毕！--------------------------------------")
# cv2.putText(frame, 'Total', (200, 60), cv2.FONT_HERSHEY_DUPLEX, 0.75, (255, 0, 0), 2)
# x2 = 200
# y2 = 100
# for i, cl in enumerate(class_nums):
# if cl > 0:
# cv2.putText(frame, LABELS[i] + "=" + str(cl), (x2, y2), cv2.FONT_HERSHEY_DUPLEX, 0.5, (255, 0, 0), 2)
# y2 = y2 + 20
cv2.putText(frame, "FPS: %f" % (fps), (int(20), int(40)), 0, 5e-3 * 200, (0, 255, 0), 3) # ！！！！！！！！！输出FPS
x2 = 180
y2 = 100
for i, cl in enumerate(counter):
if cl > 0:
cv2.putText(frame, LABELS[i] + "=" + str(cl), (x2, y2), cv2.FONT_HERSHEY_DUPLEX, 0.6, (0, 255, 255), 1)
y2 = y2 + 20
# cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
fps = (fps + (1. / (time.time() - t1))) / 2
# print("FPS = %f"%(fps))
frame_cnt += 1
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def main(yolo):
# Definition of the parameters
max_cosine_distance = 0.3
nn_budget = None
nms_max_overlap = 1.0
# Deep SORT
model_filename = 'model_data/mars-small128.pb'
encoder = gdet.create_box_encoder(model_filename, batch_size=1)
metric = nn_matching.NearestNeighborDistanceMetric("cosine",
max_cosine_distance,
nn_budget)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = True
asyncVideo_flag = False
file_path = 'IMG_3326.MOV'
dfObj = pd.DataFrame(
columns=['frame_num', 'track', 'cx', 'cy', 'w', 'h', 'track_temp'])
dfObjDTP = pd.DataFrame(columns=[
'filename', 'frame_num', 'bb1', 'bb2', 'bb3', 'bb4', 'track',
'track_temp', 'Height'
])
if asyncVideo_flag:
video_capture = VideoCaptureAsync(file_path)
else:
video_capture = cv2.VideoCapture(file_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
fps = 0.0
fps_imutils = imutils.video.FPS().start()
while True:
ret, frame = video_capture.read() # frame shape 640*480*3
if ret != True:
break
t1 = time.time()
image = Image.fromarray(frame[..., ::-1]) # bgr to rgb
boxes, confidence, classes = yolo.detect_image(image)
if tracking:
features = encoder(frame, boxes)
detections = [
Detection(bbox, confidence, cls, feature)
for bbox, confidence, cls, feature in zip(
boxes, confidence, classes, features)
]
else:
detections = [
Detection_YOLO(bbox, confidence, cls)
for bbox, confidence, cls in zip(boxes, confidence, classes)
]
# Run non-maxima suppression.
boxes = np.array([d.tlwh for d in detections])
scores = np.array([d.confidence for d in detections])
indices = preprocessing.non_max_suppression(boxes, nms_max_overlap,
scores)
detections = [detections[i] for i in indices]
if tracking:
# Call the tracker
tracker.predict()
tracker.update(detections)
for track in tracker.tracks:
if not track.is_confirmed() or track.time_since_update > 1:
continue
bbox = track.to_tlbr()
#Ini buat cropping gambar per frame
#cropped_image = frame[int(bbox[1]):int(bbox[1])+(int(bbox[3])-int(bbox[1])),int(bbox[0]):int(bbox[0])+(int(bbox[2])-int(bbox[0]))]
cropped_image = frame[int(bbox[1]):int(bbox[1]) + 256,
int(bbox[0]):int(bbox[0]) + 128]
# cropped_image = frame[2:5,6:10]
# Matiin atau comment biar ga ada box putih
# cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 255, 255), 2)
#
# cv2.putText(frame, "ID: " + str(track.track_id), (int(bbox[0]), int(bbox[1])), 0,
# 1.5e-3 * frame.shape[0], (0, 255, 0), 1)
# print(cropped_image)
dirname = "output_crop/{}".format(track.track_id)
if not os.path.exists(dirname):
os.makedirs(dirname)
if (cropped_image.size == 0):
continue
else:
writeStatus = cv2.imwrite(
"output_crop/{}/frame_{}.png".format(
track.track_id, frame_index), cropped_image)
print("output_crop/{}/frame_{}.png".format(
track.track_id, frame_index))
# Write CSV
dfObj = dfObj.append(pd.Series([
frame_index, track.track_id,
int(bbox[0]),
int(bbox[1]),
int(bbox[2]) - int(bbox[0]),
int(bbox[3]) - int(bbox[1]), track.time_since_update
],
index=dfObj.columns),
ignore_index=True)
dfObjDTP = dfObjDTP.append(pd.Series([
file_path, frame_index,
int(bbox[0]),
int(bbox[1]),
int(bbox[2]),
int(bbox[3]), track.track_id, track.time_since_update,
int(bbox[3]) - int(bbox[1])
],
index=dfObjDTP.columns),
ignore_index=True)
for det in detections:
bbox = det.to_tlbr()
score = "%.2f" % round(det.confidence * 100, 2) + "%"
#Matiin atau comment biar ga ada box putih di crop image
# cv2.rectangle(frame, (int(bbox[0]), int(bbox[1])), (int(bbox[2]), int(bbox[3])), (255, 0, 0), 2)
# if len(classes) > 0:
# cls = det.cls
# cv2.putText(frame, str(cls) + " " + score, (int(bbox[0]), int(bbox[3])), 0,
# 1.5e-3 * frame.shape[0], (0, 255, 0), 1)
cv2.imshow('', frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
dfObj = dfObj.sort_values(["track", "frame_num"], ascending=(True, True))
dfObj.to_csv(r'result_temp.csv', index=False)
dfObjDTP = dfObjDTP.sort_values(["track", "frame_num"],
ascending=(True, True))
dfObjDTP.to_csv(r'result_temp_dtp.csv', index=False)
convert_to_final()
cv2.destroyAllWindows()
def run_video(self):
# init for classify module
clf_model = None
clf_labels = None
encoder = gdet.create_box_encoder(self.cfg.DEEPSORT.MODEL,
batch_size=4)
metric = nn_matching.NearestNeighborDistanceMetric(
"cosine", self.cfg.DEEPSORT.MAX_COSINE_DISTANCE,
self.cfg.DEEPSORT.NN_BUDGET)
tracker = Tracker(self.cfg, metric)
tracking = True
writeVideo_flag = self.args.out_video
asyncVideo_flag = False
list_classes = ['loai_1', 'loai_2', 'loai_3', 'loai_4']
arr_cnt_class = np.zeros((len(list_classes), self.number_MOI),
dtype=int)
fps = 0.0
fps_imutils = imutils.video.FPS().start()
counted_obj = []
count_frame = 0
objs_dict = {}
if asyncVideo_flag:
video_capture = VideoCaptureAsync(self.video_path)
else:
video_capture = cv2.VideoCapture(self.video_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
output_camname = 'output' + '_' + self.video_name + '.avi'
out = cv2.VideoWriter(output_camname, fourcc, 10, (1280, 720))
frame_index = -1
while True:
count_frame += 1
print('frame processing .......: ', count_frame)
ret, frame = video_capture.read()
if ret != True:
break
frame_info = self.video_name + "_" + str(count_frame - 1)
t1 = time.time()
# frame = cv2.flip(frame, -1)
_frame = self.process(frame, count_frame, frame_info, encoder,
tracking, tracker, objs_dict, counted_obj,
arr_cnt_class, clf_model, clf_labels)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(_frame)
frame_index = frame_index + 1
# visualize
if self.args.visualize:
_frame = imutils.resize(_frame, width=1000)
cv2.imshow("Final result", _frame)
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()
def run_video(self):
# init for classify module
clf_model = None
clf_labels = None
if self.use_classify:
clf_model, clf_labels = mobileNet.load_model_clf(self.cfg)
encoder = gdet.create_box_encoder(self.cfg.DEEPSORT.MODEL,
batch_size=4)
metric = nn_matching.NearestNeighborDistanceMetric(
"cosine", self.cfg.DEEPSORT.MAX_COSINE_DISTANCE,
self.cfg.DEEPSORT.NN_BUDGET)
tracker = Tracker(metric)
tracking = True
writeVideo_flag = True
asyncVideo_flag = False
list_classes = ['loai_1', 'loai_2', 'loai_3', 'loai_4']
arr_cnt_class = np.zeros((len(list_classes), self.number_MOI),
dtype=int)
fps = 0.0
fps_imutils = imutils.video.FPS().start()
counted_obj = []
count_frame = 0
objs_dict = {}
# file_path = 'data/demo.MOV'
if asyncVideo_flag:
video_capture = VideoCaptureAsync(self.video_path)
else:
video_capture = cv2.VideoCapture(self.video_path)
if asyncVideo_flag:
video_capture.start()
if writeVideo_flag:
if asyncVideo_flag:
w = int(video_capture.cap.get(3))
h = int(video_capture.cap.get(4))
else:
w = int(video_capture.get(3))
h = int(video_capture.get(4))
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('output_yolov4.avi', fourcc, 30, (w, h))
frame_index = -1
while True:
count_frame += 1
ret, frame = video_capture.read()
if ret != True:
break
t1 = time.time()
# frame = cv2.flip(frame, -1)
_frame = frame
_frame = MOI.config_cam(_frame, self.cfg)
# draw board
ROI_board = np.zeros((150, 170, 3), np.int)
_frame[0:150, 0:170] = ROI_board
_frame, list_col = init_board(_frame, self.number_MOI)
_frame_height, _frame_width = _frame.shape[:2]
cropped_frame = frame
# cv2.rectangle(_frame, (int(frame_width*0), int(_frame_height*0.1)), (int(_frame_width*0.98), int(_frame_height*0.98)), (255, 0, 0), 2)
print("[INFO] Detecting.....")
detections, detections_in_ROI = self.run_detection(
cropped_frame, encoder, tracking, count_frame)
print("[INFO] Tracking....")
_, objs_dict = self.draw_tracking(cropped_frame, tracker, tracking,
detections_in_ROI, count_frame,
objs_dict)
print("[INFO] Counting....")
_frame, arr_cnt_class, vehicles_detection_list = self.counting(count_frame, cropped_frame, _frame, \
objs_dict, counted_obj,
arr_cnt_class, clf_model, clf_labels)
# delete counted id
for track in tracker.tracks:
if int(track.track_id) in counted_obj:
track.delete()
# write result to txt
with open(self.result_filename, 'a+') as result_file:
for frame_id, movement_id, vehicle_class_id in vehicles_detection_list:
result_file.write('{} {} {} {}\n'.format(
self.video_name, frame_id, movement_id,
vehicle_class_id))
# write number to scoreboard
_frame = write_board(_frame, arr_cnt_class, list_col,
self.number_MOI)
# visualize
if self.args.visualize:
_frame = imutils.resize(_frame, width=1000)
cv2.imshow("Final result", _frame)
if writeVideo_flag: # and not asyncVideo_flag:
# save a frame
out.write(_frame)
frame_index = frame_index + 1
fps_imutils.update()
if not asyncVideo_flag:
fps = (fps + (1. / (time.time() - t1))) / 2
print("FPS = %f" % (fps))
# Press Q to stop!
if cv2.waitKey(1) & 0xFF == ord('q'):
break
fps_imutils.stop()
print('imutils FPS: {}'.format(fps_imutils.fps()))
if asyncVideo_flag:
video_capture.stop()
else:
video_capture.release()
if writeVideo_flag:
out.release()
cv2.destroyAllWindows()