def csrt_create(bounding_box, frame):
#print(bounding_box)
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, tuple(bounding_box))
blob = Blob(bounding_box, tracker)
#print(blob)
return blob
DETECTION_FRAME_RATE = 48
MAX_CONSECUTIVE_TRACKING_FAILURES = 15
def get_bounding_boxes(_frame):
fullbody_cascade = cv2.CascadeClassifier('./HaarCascades/fullbody.xml')
gray = cv2.cvtColor(_frame, cv2.COLOR_BGR2GRAY)
_bounding_boxes = fullbody_cascade.detectMultiScale(gray)
return _bounding_boxes
# initialize trackers and create new blobs
_, frame = cap.read()
initial_bboxes = get_bounding_boxes(frame)
for box in initial_bboxes:
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, tuple(box))
_blob = Blob(box, tracker)
blobs[blob_id] = _blob
while True:
if cap.get(cv2.CAP_PROP_POS_FRAMES) + 1 < cap.get(cv2.CAP_PROP_FRAME_COUNT):
_, frame = cap.read()
# update trackers
for _id, blob in list(blobs.items()):
success, box = blob.tracker.update(frame)
if success:
blob.num_consecutive_tracking_failures = 0
blob.update(box)
# draw and label bounding boxes
(x, y, w, h) = [int(v) for v in box]
def csrt_create(bounding_box, frame):
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, tuple(bounding_box[0:4]))
blob = Blob(bounding_box, tracker)
return blob
def csrt_create(bounding_box, frame, vehicle_type):
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, tuple(bounding_box))
blob = Blob(bounding_box, tracker, vehicle_type)
return blob
def csrt_create(bounding_box, vehicle_type, type_confidence, frame):
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, tuple(bounding_box))
blob = Blob(bounding_box, vehicle_type, type_confidence, tracker)
return blob
def kcf_create(bounding_box, frame):
tracker = cv2.TrackerKCF_create()
tracker.init(frame, tuple(bounding_box))
blob = Blob(bounding_box, tracker)
return blob