def main():
if len(sys.argv) != 2:
print('incorrect format of execution')
return
path_to_images = sys.argv[1]
if not os.path.exists(path_to_images):
print('incorrect path to images')
return
people_with_glasses = []
input_shape = (150, 150, 3)
model = create_convnet(input_shape)
model.load_weights('./models/convnet.h5')
print('loading model...')
print('start detection')
for f in os.listdir(path_to_images):
f_full = os.path.join(path_to_images, f)
image = detect_face(f_full)
if image is None:
print("detector couldn't find a face: %s" % f)
continue
result = model.predict(np.array([image]), batch_size=16)
if result[0] > 0.5:
people_with_glasses.append(f_full)
print('done!')
print("detected %d image (people with glasses)" % len(people_with_glasses))
people_with_glasses.sort()
if people_with_glasses:
print('\n'.join(people_with_glasses))
with open('results.txt', 'w') as f:
for path in people_with_glasses:
f.write(path + '\n')
def add_frame(self, image):
mask = utils.segment(image)
face, foundFace = utils.detect_face(image)
mask = utils.eliminate_face(face, foundFace, mask)
if self.__using_stabilization:
mask = utils.stabilize(foundFace, self.__no_of_frames + 1, image,
face, mask)
hand, hand_contour = utils.get_my_hand(mask, True)
hand_pose, direction = 'None', 'None'
if hand_contour is not None:
motion_detected = self.__motion.get_hand_motion(hand_contour)
else:
return
if not motion_detected:
hand_pose = recognise_hand_pose(
hand,
directly_from_hand=True,
model_path='Models/silatra_gesture_signs.sav')
else:
direction = motion_detected
self.__observations.append((hand_pose, direction))
self.__no_of_frames += 1
def read_image(filename):
img = cv2.imread(filename)
rectangles = utils.detect_face(img)
faces = []
for rect in rectangles:
landmarks = utils.align_face(img, rect)
faces.append(((rect.left(), rect.top(), rect.right(),
rect.bottom()), landmarks))
return img, faces
def detect_face_and_save(input_folder, output_folder, filename):
output_path = os.path.join(output_folder, filename)
if os.path.exists(output_path):
return
path_to_image = os.path.join(input_folder, filename)
image_array = detect_face(path_to_image)
if image_array is None:
return 0
image = Image.fromarray(image_array)
image.save(output_path)
return 1
def load_image(self, filename):
img = cv2.imread(filename)
rectangles = utils.detect_face(img)
if not rectangles:
return False
rect = rectangles[0]
landmarks = utils.align_face(img, rect)
self.img_rect = (rect.left(), rect.top(), rect.width(), rect.height())
self.clear_sequence()
self.add_to_sequence(img, landmarks)
return True
def _preprocess(color_path, depth_path, label, save_path, length,
img_size):
if not (color_path.exists() and depth_path.exists()):
print('Sample not found, skipped. {}'.format(
color_path.parents[0]))
return
# read color, depth frames
color_video = dataio.read_video(color_path)
depth_video = dataio.read_video(depth_path)
T, H, W, C = color_video.shape
if T < length:
return
# crop to be a square (H, H) video,
tr_y, tr_x, bl_y, bl_x = utils.detect_face(color_video)
if tr_y == -1:
return
center_x = (tr_x - bl_x) // 2 + bl_x
left_x = max(center_x - (H // 2), 0)
color_video = color_video[:, :, left_x:left_x + H]
depth_video = depth_video[:, :, left_x:left_x + H]
# resize
color_video = [
imresize(img, (img_size, img_size)) for img in color_video
]
depth_video = [
imresize(img, (img_size, img_size), 'nearest')
for img in depth_video
]
color_video, depth_video = np.stack(color_video), np.stack(depth_video)
depth_video = depth_video[..., 0] # save as grayscale image
# save
name = "{}_{}_{}".format(color_path.parents[0].name,
color_path.name[2:7], label)
dataio.save_video_as_images(color_video, save_path / name / 'color')
dataio.save_video_as_images(depth_video, save_path / name / 'depth')
(save_path / 'color').mkdir(parents=True, exist_ok=True)
(save_path / 'depth').mkdir(parents=True, exist_ok=True)
dataio.write_video(color_video, save_path / 'color' / (name + ".mp4"))
dataio.write_video(depth_video, save_path / 'depth' / (name + ".mp4"))
return [name, T]
def detect(self, img, landmarks=False):
"""
:param img:
:param landmarks:
:return:
"""
with torch.no_grad():
batch_boxes, batch_points = detect_face(img, self.min_face_size,
self.pnet, self.rnet,
self.onet, self.thresholds,
self.factor, self.device)
boxes, conf, points = [], [], []
for box, point in zip(batch_boxes, batch_points):
box = np.array(box) # (len(box), 5)
point = np.array(point) # (len(box), 5)
if len(box) == 0:
boxes.append(None)
conf.append(None)
points.append(None)
elif self.select_largest:
# 按照框的面积从大到小排序索引
box_order = np.argsort(
(box[:, 2] - box[:, 0]) * (box[:, 3] - box[:, 1]))[::-1]
box = box[box_order]
point = point[box_order]
boxes.append(box[:, :4])
conf.append(box[:, 4])
points.append(point)
pass
else:
boxes.append(box[:, :4])
conf.append(box[:, 4])
points.append(point)
boxes = np.array(boxes)
conf = np.array(conf)
points = np.array(points)
if not isinstance(img, (list, tuple)):
boxes = boxes[0]
conf = conf[0]
points = points[0]
if landmarks:
return boxes, conf, points
return boxes, conf
def recognise_hand_pose(image,
directly_from_hand=False,
model_path='Models/silatra_digits_and_letters.sav',
using_stabilization=False,
no_of_frames=1):
'''
### SiLaTra Hand Pose Recognition
Provides classification for input hand pose image.
Inputs: (a) Mandatory Parameter - Image for which Hand Pose Classification is to be performed.
(b) Optional Parameters (Use them only if you understand them):
(1) directly_from_hand - boolean - Set this to true if you are passing already cropped hand region in `image` parameter.
(2) model_path - String - If an alternate model is to be used, pass the path of its .sav file.
(3) using_stabilization - boolean - If you intend to use Object stabilization, set this to True. Only use this option if you are classifying hand poses from a continuous feed, else its useless.
(4) no_of_frames - Integer - ONLY TO BE USED IF using_stabilization IS True, pass the number of the frame from the continuous feed you are processing.
'''
import pickle
from sklearn.neighbors import KNeighborsClassifier
if not directly_from_hand:
mask = utils.segment(image)
face, foundFace = utils.detect_face(image)
mask = utils.eliminate_face(face, foundFace, mask)
if using_stabilization:
mask = utils.stabilize(foundFace, no_of_frames, image, face, mask)
hand = utils.get_my_hand(mask)
if hand is False: return 'No hand pose in image'
features = utils.extract_features(hand)
else:
features = utils.extract_features(image)
classifier = pickle.load(open(model_path, 'rb'))
hand_pose = classifier.predict([features])[0]
return hand_pose
def extract_oneface(image, marigin=16):
# detecting faces
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
h, w, c = image.shape
total_boxes, points = detect_face(image, 20, pnet, rnet, onet,
[0.6, 0.7, 0.7], 0.709)
for idx, (bounding_box, keypoints) in enumerate(zip(total_boxes,
points.T)):
bounding_boxes = {
'box': [
int(bounding_box[0]),
int(bounding_box[1]),
int(bounding_box[2] - bounding_box[0]),
int(bounding_box[3] - bounding_box[1])
],
'confidence':
bounding_box[-1],
'keypoints': {
'left_eye': (int(keypoints[0]), int(keypoints[5])),
'right_eye': (int(keypoints[1]), int(keypoints[6])),
'nose': (int(keypoints[2]), int(keypoints[7])),
'mouth_left': (int(keypoints[3]), int(keypoints[8])),
'mouth_right': (int(keypoints[4]), int(keypoints[9])),
}
}
bounding_box = bounding_boxes['box']
keypoints = bounding_boxes['keypoints']
# align face and extract it out
align_image = align_face(image, keypoints)
align_image = cv2.cvtColor(align_image, cv2.COLOR_RGB2BGR)
xmin = max(bounding_box[0] - marigin, 0)
ymin = max(bounding_box[1] - marigin, 0)
xmax = min(bounding_box[0] + bounding_box[2] + marigin, w)
ymax = min(bounding_box[1] + bounding_box[3] + marigin, h)
crop_image = align_image[ymin:ymax, xmin:xmax, :]
# "just need only one face"
return crop_image
def recognise_hand_pose(image, directly_from_hand=False, model_path='Models/digits_and_letters.sav', using_stabilization=False, no_of_frames=1):
import pickle
from sklearn.neighbors import KNeighborsClassifier
if not directly_from_hand:
mask = utils.segment(image)
face, foundFace = utils.detect_face(image)
mask = utils.eliminate_face(face, foundFace, mask)
if using_stabilization: mask = utils.stabilize(foundFace, no_of_frames, image, face, mask)
hand = utils.get_my_hand(mask)
if hand is False: return 'No hand pose in image'
features = utils.extract_features(hand)
else: features = utils.extract_features(image)
classifier = pickle.load(open(model_path, 'rb'))
hand_pose = classifier.predict([features])[0]
return hand_pose
def __init__(self, args):
self.args = args
self.detector_faces, self.detector_landmarks = detect_face(args.model)
self.sunglasses, self.points = load_assets(args.image)
self.sun_h, self.sun_w, _ = self.sunglasses.shape
def run_on_image(net, height_size, cpu, track, smooth, img, stride,
upsample_ratio, num_keypoints, threshold):
global previous_poses
orig_img = img.copy()
heatmaps, pafs, scale, pad = infer_fast(net, img, height_size, stride,
upsample_ratio, cpu)
score = 0
total_keypoints_num = 0
all_keypoints_by_type = []
for kpt_idx in range(num_keypoints): # 19th for bg
total_keypoints_num += extract_keypoints(heatmaps[:, :, kpt_idx],
all_keypoints_by_type,
total_keypoints_num)
pose_entries, all_keypoints = group_keypoints(all_keypoints_by_type,
pafs,
demo=True)
for kpt_id in range(all_keypoints.shape[0]):
all_keypoints[kpt_id,
0] = (all_keypoints[kpt_id, 0] * stride / upsample_ratio
- pad[1]) / scale
all_keypoints[kpt_id,
1] = (all_keypoints[kpt_id, 1] * stride / upsample_ratio
- pad[0]) / scale
current_poses = []
for n in range(len(pose_entries)):
if len(pose_entries[n]) == 0:
continue
pose_keypoints = np.ones((num_keypoints, 2), dtype=np.int32) * -1
for kpt_id in range(num_keypoints):
if pose_entries[n][kpt_id] != -1.0: # keypoint was found
pose_keypoints[kpt_id, 0] = int(
all_keypoints[int(pose_entries[n][kpt_id]), 0])
pose_keypoints[kpt_id, 1] = int(
all_keypoints[int(pose_entries[n][kpt_id]), 1])
pose = Pose(pose_keypoints, pose_entries[n][18])
current_poses.append(pose)
if track:
track_poses(previous_poses, current_poses, smooth=smooth)
previous_poses = current_poses
for pose in current_poses:
pose.draw(img)
img = cv2.addWeighted(orig_img, 0.6, img, 0.4, 0)
for pose in current_poses:
# cv2.rectangle(img, (pose.bbox[0], pose.bbox[1]),
r_hand_center, r_hand_width, l_hand_center, l_hand_width, = detect_hand(
pose)
if -1 not in r_hand_center:
cv2.circle(img, (r_hand_center[0], r_hand_center[1]), 5,
(255, 0, 0), 5)
cv2.rectangle(img, (r_hand_center[0] - r_hand_width,
r_hand_center[1] - r_hand_width),
(r_hand_center[0] + r_hand_width,
r_hand_center[1] + r_hand_width), (0, 255, 255))
if -1 not in l_hand_center:
cv2.circle(img, (l_hand_center[0], l_hand_center[1]), 5,
(255, 0, 0), 5)
cv2.rectangle(img, (l_hand_center[0] - l_hand_width,
l_hand_center[1] - l_hand_width),
(l_hand_center[0] + l_hand_width,
l_hand_center[1] + l_hand_width), (0, 255, 255))
face_center, face_width = detect_face(pose)
if -1 not in face_center:
cv2.rectangle(
img,
(face_center[0] - face_width, face_center[1] - face_width),
(face_center[0] + face_width, face_center[1] + face_width),
(0, 0, 255))
# (pose.bbox[0] + pose.bbox[2], pose.bbox[1] + pose.bbox[3]), (0, 255, 0))
if track:
cv2.putText(img, 'id: {}'.format(pose.id),
(face_center[0] - face_width,
face_center[1] - face_width - 16),
cv2.FONT_HERSHEY_COMPLEX, 0.5, (0, 0, 255))
if -1 not in r_hand_center:
x, y, h, w, score = detect_touch(face_center, face_width,
r_hand_center, r_hand_width)
if h != 0:
cv2.rectangle(img, (x, y), (x + h, y + w), (255, 0, 255))
cv2.putText(img, f'Score: {score:0.2f}', (x, y - 16),
cv2.FONT_HERSHEY_COMPLEX, 0.5, (255, 255, 0))
if -1 not in l_hand_center:
x, y, h, w, score = detect_touch(face_center, face_width,
l_hand_center, l_hand_width)
if h != 0:
cv2.rectangle(img, (x, y), (x + h, y + w), (255, 0, 255))
cv2.putText(img, f'Score: {score:0.2f}', (x, y - 16),
cv2.FONT_HERSHEY_COMPLEX, 0.5, (255, 255, 0))
cv2.imshow('Lightweight Human Pose Estimation Python Demo', img)
delay = 1
detect = False
key = cv2.waitKey(delay)
if key == 27: # esc
return
elif key == 112: # 'p'
if delay == 33:
delay = 0
else:
delay = 33
return score > threshold
cv2.namedWindow("detecting face")
cap = cv2.VideoCapture(0)
while(cap.isOpened()):
ret,image = cap.read()
if ret == True:
# resize image
image_h, image_w, _ = image.shape
new_h, new_w = int(0.5*image_h), int(0.5*image_w)
image = cv2.resize(image, (new_w, new_h))
org_image = image.copy()
# detecting faces
# t1 = time.time()
image = cv2.cvtColor(image ,cv2.COLOR_BGR2RGB)
total_boxes, points = detect_face(image, 20, pnet, rnet, onet, [0.6, 0.7, 0.7], 0.709)
# t2 = time.time()
# print("time: %.2fms" %((t2-t1)*1000))
for idx, (bounding_box, keypoints) in enumerate(zip(total_boxes, points.T)):
bounding_boxes = {
'box': [int(bounding_box[0]), int(bounding_box[1]),
int(bounding_box[2]-bounding_box[0]), int(bounding_box[3]-bounding_box[1])],
'confidence': bounding_box[-1],
'keypoints': {
'left_eye': (int(keypoints[0]), int(keypoints[5])),
'right_eye': (int(keypoints[1]), int(keypoints[6])),
'nose': (int(keypoints[2]), int(keypoints[7])),
'mouth_left': (int(keypoints[3]), int(keypoints[8])),
'mouth_right': (int(keypoints[4]), int(keypoints[9])),
}
print("Wrong input format, example : ")
print("python test.py input_image_path")
exit()
path_i = args[1]
G = Generator(32)
GR = Generator_R(32)
G.load_state_dict(
torch.load("Checkpoints/G_22.pt", map_location=torch.device('cpu')))
GR.load_state_dict(
torch.load("Checkpoints/GR_7.pt", map_location=torch.device('cpu')))
img = plt.imread(path_i)
faces = detect_face(img)
if len(faces) == 0:
print("No faces detected")
exit()
resz = cv2.resize(faces[0], (100, 100))
plt.imsave("out_ld.png", resz)
resz = resz.reshape(1, 100, 100, 3)
resz = np.transpose(resz, (0, 3, 1, 2))
resz = torch.from_numpy(resz)
resz = resz.float()
inp = scale(resz)
out1 = infer(G, inp)
out2 = infer(GR, inp)
#! /usr/bin/env python
import cv2
from swap import swap_images
from utils import landmark_detection, detect_face, readPoints
if __name__ == '__main__':
# Read images
cap = cv2.VideoCapture(0)
while (True):
# Capture frame-by-frame
ret, img = cap.read()
gray, rects = detect_face(img)
if len(rects) == 1:
points1, _ = landmark_detection(gray, rects)
filename2 = 'donald_trump.jpg'
img2 = cv2.imread(filename2)
points2 = readPoints(filename2 + '.txt')
output = swap_images(img2, img, points2, points1)
cv2.imshow("Face Swapped", output)
elif len(rects) == 2:
points1, points2 = landmark_detection(gray, rects)
output = swap_images(img, img, points2, points1)
output = swap_images(img, output, points1, points2)
cv2.imshow("Face Swapped", output)
else:
cv2.imshow("Face Swapped", img)
if cv2.waitKey(1) & 0xFF == ord('q'):
