def face_detect(self, images):
detector = face_detection.FaceAlignment(
face_detection.LandmarksType._2D,
flip_input=False,
face_detector=self.args.face_detector)
batch_size = self.args.face_det_batch_size
while 1:
predictions = []
try:
for i in tqdm(range(0, len(images), batch_size)):
predictions.extend(
detector.get_detections_for_batch(
np.array(images[i:i + batch_size])))
except RuntimeError:
if batch_size == 1:
raise RuntimeError(
'Image too big to run face detection on GPU. Please use the --resize_factor argument'
)
batch_size //= 2
print('Recovering from OOM error; New batch size: {}'.format(
batch_size))
continue
break
results = []
pady1, pady2, padx1, padx2 = self.args.pads
for rect, image in zip(predictions, images):
if rect is None:
cv2.imwrite(
'temp/faulty_frame.jpg',
image) # check this frame where the face was not detected.
raise ValueError(
'Face not detected! Ensure the video contains a face in all the frames.'
)
y1 = max(0, rect[1] - pady1)
y2 = min(image.shape[0], rect[3] + pady2)
x1 = max(0, rect[0] - padx1)
x2 = min(image.shape[1], rect[2] + padx2)
results.append([x1, y1, x2, y2])
boxes = np.array(results)
if not self.args.nosmooth:
boxes = self.get_smoothened_boxes(boxes, T=5)
results = [[image[y1:y2, x1:x2], (y1, y2, x1, x2)]
for image, (x1, y1, x2, y2) in zip(images, boxes)]
del detector
return results
def extract_bbox(self, image):
detector = face_detection.FaceAlignment(
face_detection.LandmarksType._2D,
flip_input=False,
face_detector=self.face_detector)
frame = [image]
predictions = detector.get_detections_for_image(np.array(frame))
person_num = len(predictions)
if person_num == 0:
return np.array([])
results = []
face_boxs = []
h, w, _ = image.shape
for rect in predictions:
bh = rect[3] - rect[1]
bw = rect[2] - rect[0]
cy = rect[1] + int(bh / 2)
cx = rect[0] + int(bw / 2)
margin = max(bh, bw)
y1 = max(0, cy - margin)
x1 = max(0, cx - int(0.8 * margin))
y2 = min(h, cy + margin)
x2 = min(w, cx + int(0.8 * margin))
area = (y2 - y1) * (x2 - x1)
results.append([x1, y1, x2, y2, area])
# if a person has more than one bbox, keep the largest one
# maybe greedy will be better?
sorted(results, key=lambda area: area[4], reverse=True)
results_box = [results[0]]
for i in range(1, person_num):
num = len(results_box)
add_person = True
for j in range(num):
pre_person = results_box[j]
iou = self.IOU(pre_person[0], pre_person[1], pre_person[2],
pre_person[3], pre_person[4], results[i][0],
results[i][1], results[i][2], results[i][3],
results[i][4])
if iou > 0.5:
add_person = False
break
if add_person:
results_box.append(results[i])
boxes = np.array(results_box)
return boxes
def extract_bbox(self, image):
detector = face_detection.FaceAlignment(
face_detection.LandmarksType._2D, flip_input=False)
frame = [image]
predictions = detector.get_detections_for_image(np.array(frame))
results = []
h, w, _ = image.shape
for rect in predictions:
bh = rect[3] - rect[1]
bw = rect[2] - rect[0]
cy = rect[1] + int(bh / 2)
cx = rect[0] + int(bw / 2)
margin = max(bh, bw)
y1 = max(0, cy - margin)
x1 = max(0, cx - int(0.8 * margin))
y2 = min(h, cy + margin)
x2 = min(w, cx + int(0.8 * margin))
results.append([x1, y1, x2, y2])
boxes = np.array(results)
return boxes
type=int)
parser.add_argument('--batch_size',
help='Single GPU Face detection batch size',
default=32,
type=int)
parser.add_argument("--data_root",
help="Root folder of the LRS2 dataset",
required=True)
parser.add_argument("--preprocessed_root",
help="Root folder of the preprocessed dataset",
required=True)
args = parser.parse_args()
fa = [
face_detection.FaceAlignment(face_detection.LandmarksType._2D,
flip_input=False)
]
template = 'ffmpeg -loglevel panic -y -i {} -strict -2 {}'
# template2 = 'ffmpeg -hide_banner -loglevel panic -threads 1 -y -i {} -async 1 -ac 1 -vn -acodec pcm_s16le -ar 16000 {}'
def process_video_file(vfile, args, gpu_id):
video_stream = cv2.VideoCapture(vfile)
frames = []
while 1:
still_reading, frame = video_stream.read()
if not still_reading:
video_stream.release()
break