/
demo.py
162 lines (144 loc) · 6.22 KB
/
demo.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
import argparse
import cv2
import time
import os
parser = argparse.ArgumentParser(description='Test')
parser.add_argument('--face_det', default='pig',
type=str, help='Method used to detect faces.')
parser.add_argument('--landmark_det', default='pig',
type=str, help='Method used to detect landmarks.')
parser.add_argument("--image", type=str, default=None,
help="image file to be processed.")
parser.add_argument("--video", type=str, default='test/test_eye.mp4',
help="Video file to be processed.")
parser.add_argument("--cam", type=int, default=None,
help="The webcam index.")
parser.add_argument("--save_path", type=str, default='result/result.avi',
help="The save path.")
parser.add_argument("--fps", type=int, default=15,
help="The frames per second of the video to save.")
parser.add_argument("--time_cost", type=str, default=None,
help="The path(.txt) to save time cost per frame.")
parser.add_argument("--stretchY", type=float, default=1,
help="The face boxes usually need to be stretched along axis Y,this is the stretch rate.")
args = parser.parse_args()
if __name__ == '__main__':
# Generate detectors.
face_detector = None
landmark_detector = None
if args.face_det == 'dlib':
from det_dlib import dlib_face_detector
face_detector = dlib_face_detector()
elif args.face_det == 'mtcnn':
from det_mtcnn import mtcnn_face_detector
face_detector = mtcnn_face_detector()
elif args.face_det == 'linzaer':
from det_linzaer import linzaer_face_detector
face_detector = linzaer_face_detector()
elif args.face_det == 'centerface':
from det_centerface import centerface_face_detector
face_detector = centerface_face_detector()
elif args.face_det == 'biubug':
from det_biubug import biubug_face_detector
face_detector = biubug_face_detector()
elif args.face_det == 'mobileface':
from det_mobileface import mobileface_face_detector
face_detector = mobileface_face_detector()
elif args.face_det == 'zqmtcnn':
from det_zqcnn import zqmtcnn_face_detector
face_detector = zqmtcnn_face_detector()
elif args.face_det == 'pig':
from det_pig import pig_face_detector
face_detector = pig_face_detector()
else:
print("Don't support face detector!")
exit(0)
if args.landmark_det == 'dlib':
from det_dlib import dlib_landmark_detector
landmark_detector = dlib_landmark_detector()
elif args.landmark_det == 'pfld':
from det_pfld import pfld_landmark_detector
landmark_detector = pfld_landmark_detector()
elif args.landmark_det == 'L106Net112':
from det_zqcnn import L106Net112_landmark_detector
landmark_detector = L106Net112_landmark_detector()
elif args.landmark_det == 'L106Net96':
from det_zqcnn import L106Net96_landmark_detector
landmark_detector = L106Net96_landmark_detector()
elif args.landmark_det == 'cnn':
from det_cnn import cnn_landmark_detector
landmark_detector = cnn_landmark_detector()
elif args.landmark_det == 'frda':
from det_frda import frda_landmark_detector
landmark_detector = frda_landmark_detector()
elif args.landmark_det == 'pig':
from det_pig import pig_landmark_detector
landmark_detector = pig_landmark_detector()
else:
print("Don't support landmark detector!")
exit(0)
# Make dirs.
save_path = args.save_path
if not os.path.exists(os.path.dirname(save_path)):
os.makedirs(os.path.dirname(save_path))
# Detection and visualization.
time_cost = []
if args.image is None:
video_src = args.cam if args.cam is not None else args.video
if video_src is None:
exit(0)
if args.save_path is None:
print("Save path is required!")
exit(0)
cap = cv2.VideoCapture(video_src)
fourcc = cv2.VideoWriter_fourcc(*'XVID')
# fps = cap.get(cv2.CAP_PROP_FPS)
fps = args.fps
size = (int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)), int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)))
out = cv2.VideoWriter(save_path, fourcc, fps, size)
while cap.isOpened():
ret, frame = cap.read()
# frame=cv2.flip(frame,1)
if not ret:
break
tic = time.time()
faces = face_detector.det_faces(frame) # faces detection
faces = [[face[0], face[1], face[2], int(face[3] * args.stretchY)] for face in faces] # stretch along Y
landmarks = landmark_detector.det_landmarks(frame, faces=faces) # landmarks detection
time_cost.append(str(time.time() - tic))
for face in faces:
cv2.rectangle(frame, (int(face[0]), int(face[1])),
(int(face[0] + face[2]), int(face[1] + face[3])), (0, 255, 0), 1)
for marks in landmarks:
for mark in marks:
cv2.circle(frame, (int(mark[0]), int(mark[1])), 1, (255, 0, 0), 1)
out.write(frame)
cv2.imshow('video', frame)
k = cv2.waitKey(40)
# q键退出
if k & 0xff == ord('q'):
break
cap.release()
out.release()
cv2.destroyAllWindows()
else:
if args.save_path is None:
print("Save path is required!")
exit(0)
frame = cv2.imread(args.image)
tic = time.time()
faces = face_detector.det_faces(frame)
landmarks = landmark_detector.det_landmarks(frame, faces=faces)
time_cost.append(time.time() - tic)
for face in faces:
cv2.rectangle(frame, (face[0], face[1]), (face[0] + face[2], face[1] + face[3]), (0, 255, 0), 1)
for marks in landmarks:
for mark in marks:
cv2.circle(frame, (mark[0], mark[1]), 1, (255, 0, 0), 1)
cv2.imshow('image', frame)
cv2.waitKey()
cv2.imwrite(save_path, frame)
if args.time_cost is not None:
fw = open(args.time_cost, 'w')
fw.write('\n'.join(time_cost))
fw.close()