def compute_map_gt_det(det_file, gt_file):
det_list = read_detections(det_file)
gt_list = read_detections(gt_file)
frames = []
for i in range(0, len(det_list)):
frame = Frame(i)
frame.detections = det_list[i]
frame.ground_truth = gt_list[i]
frames.append(frame)
mAP = mean_average_precision(frames, ignore_classes=True)
return mAP
def create_siamese_dataset(sequence, camera):
AICity_dataset_path = '../datasets/AICity_data/train/' + sequence + '/' + camera + '/'
siamese_dataset_path = '../datasets/full_siamese_data/'
annotation_file = 'gt/gt.txt'
gt_detections = read_detections(AICity_dataset_path + annotation_file)
frames = len(gt_detections)
for frame in range(frames):
try:
frame_name = "frame{:04d}.jpg".format(frame)
img = cv2.imread(AICity_dataset_path + 'frames/' + frame_name)
print(AICity_dataset_path + 'frames/' + frame_name)
for detection in gt_detections[frame]:
xtl, ytl = detection.top_left
w = detection.width
h = detection.height
class_path = sequence + '-' + str(detection.id) + '/'
if not os.path.exists(siamese_dataset_path + class_path):
os.mkdir(siamese_dataset_path + class_path)
cropped_img = img[ytl:ytl + h, xtl:xtl + w]
cv2.imwrite(siamese_dataset_path + class_path + frame_name,
cropped_img)
except:
import code
code.interact(local=dict(globals(), **locals()))
def week2_adaptive_hsv(video: Video, debug=False) -> Iterator[Frame]:
model_mean, model_std = get_background_model(video,
int(2141 * 0.25),
total_frames=int(2141 * 0.25),
pixel_value=PixelValue.HSV)
ground_truth = read_detections(
'../datasets/AICity_data/train/S03/c010/gt/gt.txt')
frame_id = int(2141 * 0.25)
roi = cv2.cvtColor(
cv2.imread('../datasets/AICity_data/train/S03/c010/roi.jpg'),
cv2.COLOR_BGR2GRAY)
for im, mask in gaussian_model_adaptive(video,
int(2141 * 0.25),
model_mean,
model_std,
total_frames=int(2141 * 0.10),
pixel_value=PixelValue.HSV,
alpha=1.75,
rho=0.01):
mask = mask & roi
if debug:
cv2.imshow('f', mask)
cv2.waitKey()
mask = opening(mask, 7)
if debug:
cv2.imshow('f', mask)
cv2.waitKey()
mask = closing(mask, 35)
if debug:
cv2.imshow('f', mask)
cv2.waitKey()
mask, detections = find_boxes(mask)
frame = Frame(frame_id)
frame.detections = detections
frame.ground_truth = ground_truth[frame_id]
if debug:
mask2 = cv2.cvtColor(mask, cv2.COLOR_GRAY2BGR)
for detection in detections:
cv2.rectangle(
mask2,
(int(detection.top_left[1]), int(detection.top_left[0])),
(int(detection.get_bottom_right()[1]),
int(detection.get_bottom_right()[0])), (0, 255, 0), 5)
for gt in ground_truth[frame_id]:
cv2.rectangle(mask2,
(int(gt.top_left[1]), int(gt.top_left[0])),
(int(gt.get_bottom_right()[1]),
int(gt.get_bottom_right()[0])), (255, 0, 0), 5)
cv2.imshow('f', mask2)
cv2.waitKey()
yield im, mask, frame
frame_id += 1
def get_frames(self) -> Iterable[Frame]:
video = cv2.VideoCapture(os.path.join(self.video_path, "vdo.avi"))
full_detections = read_detections(
os.path.join(self.video_path, "det/det_yolo3.txt"))
if os.path.exists(os.path.join(self.video_path, "gt/gt.txt")):
full_ground_truth = read_detections(
os.path.join(self.video_path, "gt/gt.txt"))
else:
full_ground_truth = []
count = 0
while video.isOpened():
ret, frame = video.read()
if ret:
det = full_detections[count] if count < len(
full_detections) else []
gt = full_ground_truth[count] if count < len(
full_ground_truth) else []
yield Frame(count, det, gt, frame)
count += 1
else:
video.release()
def main():
start_frame = 1440
end_frame = 1789
gt = read_annotations('../annotations', start_frame, end_frame)
alg = 'mask_rcnn'
detections = read_detections('../datasets/AICity_data/train/S03/c010/det/det_{0}.txt'.format(alg))
kalman = KalmanTracking()
for i in range(start_frame, end_frame):
f = Frame(i)
f.detections = detections[i]
f.ground_truth = gt[i - start_frame]
kalman(f)
print(seq(f.detections).map(lambda d: d.id).to_list())
def week2_soa(video: Video, debug=False) -> Iterator[Frame]:
th = 150
frame_id = 0
fgbg = cv.createBackgroundSubtractorMOG2()
ground_truth = read_detections(
'../datasets/AICity_data/train/S03/c010/gt/gt.txt')
roi = cv.cvtColor(
cv.imread('../datasets/AICity_data/train/S03/c010/roi.jpg'),
cv.COLOR_BGR2GRAY)
for im in tqdm(video.get_frames(),
total=2141,
file=sys.stdout,
desc='Training model...'):
mask = fgbg.apply(im)
mask[mask < th] = 0
mask.astype(np.uint8) * 255
mask = mask & roi
mask = opening(mask, 5)
# cv.imshow('f', mask)
# cv.waitKey()
mask = closing(mask, 25)
# cv.imshow('f', mask)
# cv.waitKey()
mask, detections = find_boxes(mask)
frame = Frame(frame_id)
frame.detections = detections
frame.ground_truth = ground_truth[frame_id]
frame_id += 1
yield im, mask, frame
def main():
video = Video("../datasets/AICity_data/train/S03/c010/vdo.avi")
gt = read_annotations('../annotations', start_frame, end_frame)
"""
DETECTIONS
"""
det_algs = ['yolo3', 'mask_rcnn', 'ssd512']
for alg in det_algs:
detections = read_detections(
'../datasets/AICity_data/train/S03/c010/det/det_{0}.txt'.format(
alg))
detections = detections[start_frame:end_frame + 1]
frames = []
# roi = cv2.imread('../datasets/AICity_data/train/S03/c010/roi.jpg')
for im, f in seq(video.get_frames(
start_frame_number=start_frame)).take(end_frame - start_frame +
1):
f.ground_truth = gt[f.id]
f.detections = detections[f.id]
frames.append(f)
if make_video:
make_video_frame(im, f, frames)
iou_over_time(frames)
mAP = mean_average_precision(frames)
print(alg, " mAP:", mAP)
"""
DETECTIONS FROM ALTERED GROUND TRUTH
"""
frames = []
for im, f in seq(video.get_frames()).take(end_frame - start_frame + 1):
f.ground_truth = gt[f.id]
f.detections = alter_detections(f.ground_truth)
frames.append(f)
if make_video:
make_video_frame(im, f, frames)
iou_over_time(frames)
mAP = mean_average_precision(frames)
print('Random alteration', " mAP:", mAP)
"""
OPTICAL FLOW
"""
of_det_1 = read_optical_flow(
'../datasets/optical_flow/detection/LKflow_000045_10.png')
of_det_2 = read_optical_flow(
'../datasets/optical_flow/detection/LKflow_000157_10.png')
of_gt_1 = read_optical_flow('../datasets/optical_flow/gt/000045_10.png')
of_gt_2 = read_optical_flow('../datasets/optical_flow/gt/000157_10.png')
img_1 = cv2.imread('../datasets/optical_flow/img/000045_10.png')
img_2 = cv2.imread('../datasets/optical_flow/img/000157_10.png')
msen_of = msen(of_det_2, of_gt_2)
pepn_of = pepn(of_det_2, of_gt_2)
print(msen_of, pepn_of)
show_optical_flow(of_gt_1)
show_optical_flow_arrows(img_1, of_gt_1)
msen_45 = msen(of_det_1, of_gt_1, plot=True)
pepn_45 = pepn(of_det_1, of_gt_1)
print("Sequence 045: MSEN", msen_45, "PEPN", pepn_45)
msen_157 = msen(of_det_2, of_gt_2, plot=True)
pepn_157 = pepn(of_det_2, of_gt_2)
print("Sequence 157: MSEN", msen_157, "PEPN", pepn_157)
show_optical_flow(of_gt_1)