def main():
parser = argparse.ArgumentParser()
parser.add_argument('--model', help='model path', default=default_model)
parser.add_argument('--labels',
help='label file path',
default=default_labels)
parser.add_argument('--threshold',
type=float,
default=0.2,
help='class score threshold')
parser.add_argument('--use_tracker',
type=bool,
default=False,
help='use an object tracker')
args = parser.parse_args()
print('Loading %s with %s labels.' % (args.model, args.labels))
config = vot.ObjectTrackingConfig(
score_threshold=args.threshold,
tracker=vot.Tracker.BASIC if args.use_tracker else vot.Tracker.NONE)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
fps_calculator = utils.FpsCalculator()
cap = cv2.VideoCapture(0)
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Converts image to PIL Image.
pil_im = Image.fromarray(cv2.cvtColor(frame,
cv2.COLOR_BGR2RGB)).resize(
(input_size.width,
input_size.height))
# Grabs current millisecond for timestamp.
timestamp = current_milli_time()
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, pil_im, annotations):
frame = utils.render_bbox(frame, annotations)
# Calculate FPS, then visualize it.
fps, latency = fps_calculator.measure()
frame = cv2.putText(frame, '{} fps'.format(fps), (0, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
frame = cv2.putText(frame, '{} ms'.format(latency), (0, 40),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
def main():
default_video = 'data/traffic_frames.mp4'
default_model = 'data/traffic_model_edgetpu.tflite'
default_labels = 'data/traffic_label_map.pbtxt'
parser = argparse.ArgumentParser()
parser.add_argument('--model', help='model path', default=default_model)
parser.add_argument('--labels',
help='label file path',
default=default_labels)
parser.add_argument('--input_video',
help='input video file path',
default=default_video)
parser.add_argument('--output_video',
help='output video file path',
default='')
parser.add_argument('--threshold',
type=float,
default=0.2,
help='class score threshold')
parser.add_argument('--use_tracker',
type=bool,
default=False,
help='use an object tracker')
args = parser.parse_args()
print('Loading %s with %s labels.' % (args.model, args.labels))
config = vot.ObjectTrackingConfig(
score_threshold=args.threshold,
tracker=vot.Tracker.BASIC if args.use_tracker else vot.Tracker.NONE)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
cap = cv2.VideoCapture(args.input_video)
writer = None
if cap.isOpened() and args.output_video:
writer = cv2.VideoWriter(args.output_video,
cv2.VideoWriter_fourcc(*'mp4v'),
cap.get(cv2.CAP_PROP_FPS),
(int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))))
timestamp = 0
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Resizes frame.
resized_frame = cv2.resize(frame,
(input_size.width, input_size.height))
# Calculates current microsecond for timestamp.
timestamp = int(timestamp +
(1 / cap.get(cv2.CAP_PROP_FPS)) * 1000 * 1000)
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, resized_frame, annotations):
frame = utils.render_bbox(frame, annotations)
if writer:
writer.write(frame)
else:
cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
if writer:
writer.release()
else:
cv2.destroyAllWindows()
cap.release()
def main():
default_video = 'data/traffic_frames.mp4'
default_model = 'data/mobilenet_ssd_v2_coco_quant_postprocess_edgetpu.tflite'
default_labels = 'data/coco_labels.txt'
parser = argparse.ArgumentParser()
parser.add_argument('--model', help='model path', default=default_model)
parser.add_argument('--labels',
help='label file path',
default=default_labels)
parser.add_argument('--input_video',
help='input video file path',
default=default_video)
parser.add_argument('--output_video',
help='output video file path',
default='')
parser.add_argument('--threshold',
type=float,
default=0.2,
help='class score threshold')
parser.add_argument('--use_tracker',
type=str,
default=None,
help='use an object tracker',
choices=[None, 'mediapipe', 'sort', 'camshift'])
args = parser.parse_args()
if args.use_tracker == 'mediapipe': #For MediaPipe Object Tracker
trackerToBeUsed = vot.Tracker.BASIC
elif args.use_tracker == 'sort': #For Sort Object Tracker
trackerToBeUsed = vot.Tracker.SORT
elif args.use_tracker == 'camshift': #For CamShift object Tracker
trackerToBeUsed = vot.Tracker.FAST_INACCURATE
else:
trackerToBeUsed = vot.Tracker.NONE
print('Loading %s with %s labels.' % (args.model, args.labels))
config = vot.ObjectTrackingConfig(score_threshold=args.threshold,
tracker=trackerToBeUsed)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
fps_calculator = utils.FpsCalculator()
cap = cv2.VideoCapture(args.input_video)
writer = None
if cap.isOpened() and args.output_video:
writer = cv2.VideoWriter(args.output_video,
cv2.VideoWriter_fourcc(*'mp4v'),
cap.get(cv2.CAP_PROP_FPS),
(int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))))
timestamp = 0
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Resizes frame.
resized_frame = cv2.resize(frame,
(input_size.width, input_size.height))
# Calculates current microsecond for timestamp.
timestamp = int(timestamp +
(1 / cap.get(cv2.CAP_PROP_FPS)) * 1000 * 1000)
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, resized_frame, annotations):
frame = utils.render_bbox(frame, annotations)
# Calculate FPS, then visualize it.
fps, latency = fps_calculator.measure()
print('fps =', fps)
print('latency =', latency)
frame = cv2.putText(frame, '{} fps'.format(fps), (0, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
frame = cv2.putText(frame, '{} ms'.format(latency), (0, 40),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
if writer:
writer.write(frame)
else:
cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
if writer:
writer.release()
else:
cv2.destroyAllWindows()
cap.release()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--model', help='model path', default=default_model)
parser.add_argument('--labels',
help='label file path',
default=default_labels)
parser.add_argument('--threshold',
type=float,
default=0.2,
help='class score threshold')
parser.add_argument('--use_tracker',
type=bool,
default=False,
help='use an object tracker')
args = parser.parse_args()
print('Loading %s with %s labels.' % (args.model, args.labels))
print('Welcome to Dominadas Madafakers')
config = vot.ObjectTrackingConfig(
score_threshold=args.threshold,
tracker=vot.Tracker.BASIC if args.use_tracker else vot.Tracker.NONE)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
fps_calculator = utils.FpsCalculator()
cap = cv2.VideoCapture(0)
#My Dominada Counter
dominadaCount = 0
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Converts image to PIL Image.
pil_im = Image.fromarray(frame).convert('RGB').resize(
(input_size.width, input_size.height))
# Grabs current millisecond for timestamp.
timestamp = current_milli_time()
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, pil_im, annotations):
frame = utils.render_bbox(frame, annotations)
print("JEGF CODE")
#print("Frame:")
#print(frame)
print("annotations:")
for annotation in annotations:
#print(annotation.class_name+str(annotation.confidence_score))
if (annotation.class_name == "dominada"):
print(annotation.class_name +
str(annotation.confidence_score))
if (annotation.confidence_score > 0.50):
print("More than 50%")
if (annotation.class_name == "dominada"):
dominadaCount = dominadaCount + 1
try:
playsound(
"/home/mendel/dominadas-autoML/automl-videoOnDevice/examples/referee.mp3"
)
print("Playing a sound")
except:
print("Couldn't play a sound")
# Calculate FPS, then visualize it.
fps, latency = fps_calculator.measure()
#frame = cv2.putText(frame, '{} fps'.format(fps), (0, 20),
#cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
#frame = cv2.putText(frame, '{} ms'.format(latency), (0, 40),
#cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
frame = cv2.putText(frame, ('Dominadas:' + str(dominadaCount)),
(0, 66), cv2.FONT_HERSHEY_SIMPLEX, 1.1,
(0, 0, 255), 2)
cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
def main():
default_model = 'data/traffic_model_tftrt.pb'
default_labels = 'data/traffic_label_map.pbtxt'
parser = argparse.ArgumentParser()
parser.add_argument('--model', help='model path', default=default_model)
parser.add_argument('--labels',
help='label file path',
default=default_labels)
parser.add_argument('--threshold',
type=float,
default=0.25,
help='class score threshold')
parser.add_argument('--use_tracker',
type=bool,
default=False,
help='use an object tracker')
parser.add_argument(
'--video_device',
help='-1 for ribbon-cable camera. >= 0 for USB camera. '
'If both are plugged in, the USB camera will have the ID "1".',
type=int,
default=-1)
parser.add_argument('--video_width',
help='Input video width.',
type=int,
default=1280)
parser.add_argument('--video_height',
help='Input video height.',
type=int,
default=720)
args = parser.parse_args()
print('Loading %s with %s labels.' % (args.model, args.labels))
config = vot.ObjectTrackingConfig(score_threshold=args.threshold,
tracker=vot.Tracker.FAST_INACCURATE if
args.use_tracker else vot.Tracker.NONE)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
fps_calculator = utils.FpsCalculator()
if args.video_device >= 0:
cap = cv2.VideoCapture(
'v4l2src device=/dev/video{} ! videoconvert ! '
'videoscale method=0 add-borders=false ! '
'video/x-raw, width={}, height={}, format=RGB ! videoconvert ! '
'appsink'.format(args.video_device, args.video_width,
args.video_height), cv2.CAP_GSTREAMER)
else:
cap = cv2.VideoCapture(
'nvarguscamerasrc ! nvvidconv ! '
'video/x-raw, format=(string)BGRx ! videoconvert ! '
'videoscale method=0 add-borders=false ! '
'video/x-raw, width={}, height={}, format=RGB ! videoconvert ! '
'appsink'.format(args.video_width, args.video_height),
cv2.CAP_GSTREAMER)
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Resizes frame.
resized_frame = cv2.resize(frame,
(input_size.width, input_size.height))
# Grabs current millisecond for timestamp.
timestamp = current_milli_time()
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, resized_frame, annotations):
frame = utils.render_bbox(frame, annotations)
# Calculate FPS, then visualize it.
fps, latency = fps_calculator.measure()
frame = cv2.putText(frame, '{} fps'.format(fps), (0, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
frame = cv2.putText(frame, '{} ms'.format(latency), (0, 40),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
cv2.imshow('frame', frame)
key = cv2.waitKey(1)
if key & 0xFF == ord('q') or key == 27:
break
cap.release()
cv2.destroyAllWindows()
def main():
parser = argparse.ArgumentParser()
parser.add_argument(
'--model', help='model path', default=default_model)
parser.add_argument(
'--labels', help='label file path', default=default_labels)
parser.add_argument(
'--threshold', type=float, default=0.2, help='class score threshold')
parser.add_argument(
'--use_tracker', type=str, default=None, help='use an object tracker', choices=[None,'mediapipe','sort','camshift'])
args = parser.parse_args()
if args.use_tracker == 'mediapipe': #For MediaPipe Object Tracker
trackerToBeUsed = vot.Tracker.BASIC
elif args.use_tracker == 'sort': #For Sort Object Tracker
trackerToBeUsed = vot.Tracker.SORT
elif args.use_tracker == 'camshift': #For CamShift object Tracker
trackerToBeUsed = vot.Tracker.FAST_INACCURATE
else:
trackerToBeUsed = vot.Tracker.NONE
print('Loading %s with %s labels.' % (args.model, args.labels))
config = vot.ObjectTrackingConfig(
score_threshold=args.threshold,
tracker=trackerToBeUsed
)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
fps_calculator = utils.FpsCalculator()
cap = cv2.VideoCapture(0)
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Converts image to PIL Image.
pil_im = Image.fromarray(frame).convert('RGB').resize(
(input_size.width, input_size.height))
# Grabs current millisecond for timestamp.
timestamp = current_milli_time()
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, pil_im, annotations):
# print (annotations)
frame = utils.render_bbox(frame, annotations)
# Calculate FPS, then visualize it.
fps, latency = fps_calculator.measure()
print ('fps =',fps)
print ('latency =',latency)
frame = cv2.putText(frame, '{} fps'.format(fps), (0, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
frame = cv2.putText(frame, '{} ms'.format(latency), (0, 40),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
def main():
#default_video = '/home/mendel/dominadas-autoML/automl-videoOnDevice/data/2020-08-30_1744.mp4'
#default_video = '/home/mendel/dominadas-autoML/automl-videoOnDevice/data/2020-08-22_1537.mp4'
#default_video = '/home/mendel/dominadas-autoML/automl-videoOnDevice/data/2020-08-18_1545.mp4'
#default_video = '/home/mendel/dominadas-autoML/automl-videoOnDevice/data/2020-08-18_1743.mp4'
default_video = '/home/mendel/dominadas-autoML/automl-videoOnDevice/data/2020-09-04_1727.mp4'
default_model = 'data/traffic_model_edgetpu.tflite'
default_labels = 'data/traffic_label_map.pbtxt'
parser = argparse.ArgumentParser()
parser.add_argument('--model', help='model path', default=default_model)
parser.add_argument('--labels',
help='label file path',
default=default_labels)
parser.add_argument('--input_video',
help='input video file path',
default=default_video)
parser.add_argument('--output_video',
help='output video file path',
default='')
parser.add_argument('--threshold',
type=float,
default=0.2,
help='class score threshold')
parser.add_argument('--use_tracker',
type=bool,
default=False,
help='use an object tracker')
args = parser.parse_args()
print('Loading %s with %s labels.' % (args.model, args.labels))
config = vot.ObjectTrackingConfig(
score_threshold=args.threshold,
tracker=vot.Tracker.BASIC if args.use_tracker else vot.Tracker.NONE)
engine = vot.load(args.model, args.labels, config)
input_size = engine.input_size()
cap = cv2.VideoCapture(args.input_video)
#My Dominada Counter
dominadaCount = 0
writer = None
if cap.isOpened() and args.output_video:
writer = cv2.VideoWriter(args.output_video,
cv2.VideoWriter_fourcc(*'mp4v'),
cap.get(cv2.CAP_PROP_FPS),
(int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))))
timestamp = 0
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
# Resizes frame.
resized_frame = cv2.resize(frame,
(input_size.width, input_size.height))
# Calculates current microsecond for timestamp.
timestamp = int(timestamp +
(1 / cap.get(cv2.CAP_PROP_FPS)) * 1000 * 1000)
# Run inference engine to populate annotations array.
annotations = []
if engine.run(timestamp, resized_frame, annotations):
frame = utils.render_bbox(frame, annotations)
print("JEGF CODE")
#print("Frame:")
#print(frame)
print("annotations:")
'''
for annotation in annotations:
#print(annotation.class_name+str(annotation.confidence_score))
if(annotation.class_name=="dominada"):
print(annotation.class_name+str(annotation.confidence_score))
if(annotation.confidence_score>=.53125):
print("More than 53125%")
if(annotation.class_name=="dominada"):
dominadaCount = dominadaCount+1;
frame = cv2.putText(frame, ('Dominadas:'+str(dominadaCount)), (0, 66),cv2.FONT_HERSHEY_SIMPLEX, 1.1, (0, 0, 255), 2)
'''
if writer:
writer.write(frame)
else:
cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
if writer:
writer.release()
else:
cv2.destroyAllWindows()
cap.release()