def VideoMain(args):
TRAINED_MODEL = '../TrainedModels/MobileNetV2/' + args.TRAINED_MODEL + '/facenet.ckpt'
model.load_weights(TRAINED_MODEL).expect_partial()
if args.MODE == 'VIDEO':
cap = cv2.VideoCapture(args.video_path)
elif args.MODE == 'WEBCAM':
cap = cv2.VideoCapture("http://192.168.1.6:4747/video")
if not (cap.isOpened()):
print("Could not open video device")
while (True):
ret, frame = cap.read()
start = time.time()
image = cv2.flip(frame, 1)
boxes, points = detector.detect(image,
select_largest=args.select_largest,
proba=False,
landmarks=True)
if (boxes is not None):
for box in boxes:
cv2.rectangle(image, tuple(
(np.int32(box[0]), np.int32(box[1]))),
tuple((np.int32(box[2]), np.int32(box[3]))),
(255, 255, 255), 1)
face = extract_face(image, box, image_size=160)
face_embedding = model(np.float32(face), True)
predictions = svm_model.predict_proba(face_embedding)
best_class_idxs = np.argmax(predictions, axis=1)
best_class_probabilities = predictions[
np.arange(len(best_class_idxs)), best_class_idxs]
for i in range(len(best_class_idxs)):
print('%4d %s: %.3f' %
(i, class_names[best_class_idxs[i]],
best_class_probabilities[i]))
predict_str = '%s: %.3f' % (class_names[best_class_idxs[i]],
best_class_probabilities[i])
pos = tuple((np.int32(box[0]), np.int32(box[1])))
cv2.putText(image, predict_str, pos, FONT, 0.5,
(255, 255, 255), 1, cv2.LINE_AA)
fps_str = 'fps: %3.f' % (int(1 / (time.time() - start)))
cv2.putText(image, fps_str, (10, 50), FONT, 1, (255, 0, 0), 2,
cv2.LINE_AA)
cv2.imshow('preview', image)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
def align(self, img, select_largest=True, save_path=None):
assert img is not None
thumbnails = []
boxes, points = self.detect(img,
select_largest=select_largest,
proba=False,
landmarks=True)
if boxes is not None:
i = 0
for box, point in zip(boxes, points):
i += 1
thumbnail = detect_face.extract_face(img, box, self.image_size)
thumbnails.append(thumbnail)
if save_path is not None:
save_img(thumbnail, save_path + 'face_' + str(i) + '.jpg')
else:
return None
return thumbnails
def ImageMain(args):
TRAINED_MODEL = '../TrainedModels/MobileNetV2/' + args.TRAINED_MODEL + '/facenet.ckpt'
model.load_weights(TRAINED_MODEL).expect_partial()
image = cv2.imread(args.image_path)
if max(image.shape[0], image.shape[1]) > 900:
scale_percent = 900 / max(image.shape[0], image.shape[1])
image = img_resize(image, scale_percent)
boxes = detector.detect(img=image,
select_largest=args.select_largest,
proba=False,
landmarks=False)
if boxes is not None:
for box in boxes:
cv2.rectangle(image, tuple((np.int32(box[0]), np.int32(box[1]))),
tuple((np.int32(box[2]), np.int32(box[3]))),
(255, 255, 255), 1)
face = extract_face(image, box, image_size=160)
face_embedding = model(np.float32(face), True)
predictions = svm_model.predict_proba(face_embedding)
best_class_idxs = np.argmax(predictions, axis=1)
best_class_probabilities = predictions[
np.arange(len(best_class_idxs)), best_class_idxs]
for i in range(len(best_class_idxs)):
print('%4d %s: %.3f' % (i, class_names[best_class_idxs[i]],
best_class_probabilities[i]))
predict_str = '%s: %.3f' % (class_names[best_class_idxs[i]],
best_class_probabilities[i])
pos = tuple((np.int32(box[0]), np.int32(box[1])))
cv2.putText(image, predict_str, pos, FONT, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
cv2.imshow('preview', image)
cv2.waitKey(0)
cv2.destroyAllWindows()
def forward(self, img, save_path=None, return_prob=False, remove_eyes = False):
"""Run MTCNN face detection on a PIL image.
Arguments:
img {PIL.Image} -- A PIL image.
Keyword Arguments:
save_path {str} -- An optional save path for the cropped image. Note that when
self.prewhiten=True, although the returned tensor is prewhitened, the saved face
image is not, so it is a true representation of the face in the input image.
(default: {None})
return_prob {bool} -- Whether or not to return the detection probability.
(default: {False})
Returns:
Union[torch.Tensor, (torch.tensor, float)] -- If detected, cropped image of a
single face with dimensions 3 x image_size x image_size. Optionally, the probability
that a face was detected. If self.keep_all is True, n detected faces are returned in an
n x 3 x image_size x image_size tensor.
"""
with torch.no_grad():
boxes = detect_face(
img, self.min_face_size,
self.pnet, self.rnet, self.onet,
self.thresholds, self.factor,
self.device
)
if len(boxes) == 0:
print('Face not found')
if return_prob:
return None, [None] if self.keep_all else None
else:
return None
if self.select_largest:
boxes = boxes[
np.argsort(
(boxes[:, 2] - boxes[:, 0]) * (boxes[:, 3] - boxes[:, 1])
)[::-1]
]
if not self.keep_all:
boxes = boxes[[0]]
faces = []
probs = []
for i, box in enumerate(boxes):
face_path = save_path
if save_path is not None and i > 0:
save_name, ext = os.path.splitext(save_path)
face_path = save_name + '_' + str(i + 1) + ext
face, prob = extract_face(img, box, self.image_size, self.margin, face_path, remove_eyes)
if self.prewhiten:
face = prewhiten(face)
faces.append(face)
probs.append(prob)
if self.keep_all:
faces = torch.stack(faces)
else:
faces = faces[0]
probs = probs[0]
if return_prob:
return faces, probs
else:
return faces
def forward(self, img, save_path=None, return_prob=False):
"""Run MTCNN face detection on a PIL image or numpy array. This method performs both
detection and extraction of faces, returning tensors representing detected faces rather
than the bounding boxes. To access bounding boxes, see the MTCNN.detect() method below.
Arguments:
img {PIL.Image, np.ndarray, or list} -- A PIL image, np.ndarray, or list.
Keyword Arguments:
save_path {str} -- An optional save path for the cropped image. Note that when
self.post_process=True, although the returned tensor is post processed, the saved
face image is not, so it is a true representation of the face in the input image.
If `img` is a list of images, `save_path` should be a list of equal length.
(default: {None})
return_prob {bool} -- Whether or not to return the detection probability.
(default: {False})
Returns:
Union[torch.Tensor, tuple(torch.tensor, float)] -- If detected, cropped image of a face
with dimensions 3 x image_size x image_size. Optionally, the probability that a
face was detected. If self.keep_all is True, n detected faces are returned in an
n x 3 x image_size x image_size tensor with an optional list of detection
probabilities. If `img` is a list of images, the item(s) returned have an extra
dimension (batch) as the first dimension.
Example:
>>> from facenet_pytorch import MTCNN
>>> mtcnn = MTCNN()
>>> face_tensor, prob = mtcnn(img, save_path='face.png', return_prob=True)
"""
# Detect faces
with torch.no_grad():
batch_boxes, batch_probs = self.detect(img)
# Determine if a batch or single image was passed
batch_mode = True
if not isinstance(img,
(list, tuple)) and not (isinstance(img, np.ndarray)
and len(img.shape) == 4):
img = [img]
batch_boxes = [batch_boxes]
batch_probs = [batch_probs]
batch_mode = False
# Parse save path(s)
if save_path is not None:
if isinstance(save_path, str):
save_path = [save_path]
else:
save_path = [None for _ in range(len(img))]
# Process all bounding boxes and probabilities
faces, probs = [], []
for im, box_im, prob_im, path_im in zip(img, batch_boxes, batch_probs,
save_path):
if box_im is None:
faces.append(None)
probs.append([None] if self.keep_all else None)
continue
if not self.keep_all:
box_im = box_im[[0]]
faces_im = []
for i, box in enumerate(box_im):
face_path = path_im
if path_im is not None and i > 0:
save_name, ext = os.path.splitext(path_im)
face_path = save_name + '_' + str(i + 1) + ext
face = extract_face(im, box, self.image_size, self.margin,
face_path)
if self.post_process:
face = fixed_image_standardization(face)
faces_im.append(face)
if self.keep_all:
faces_im = torch.stack(faces_im)
else:
faces_im = faces_im[0]
prob_im = prob_im[0]
faces.append(faces_im)
probs.append(prob_im)
if not batch_mode:
faces = faces[0]
probs = probs[0]
if return_prob:
return faces, probs
else:
return faces