-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathpredict.py
161 lines (151 loc) · 6.6 KB
/
predict.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
# -------------------------------------#
# 对单张图片进行预测
# -------------------------------------#
import onnxruntime
from PIL import Image,ImageFont, ImageDraw
import numpy as np
import time
import cv2
# 8class
crop_path = r"checkpoint/data_12_23/mobilenetv2/888/mobilenetv2_1_12_23_acc=91.6275.onnx"
# 9class
path = r"checkpoint/data_3_25_all/mobilenetv2/111/mobilenetv2_224_acc=85.6154.onnx"
# num_classes = 9
num_classes = 8
onnx_session = onnxruntime.InferenceSession(path, None)
labels_8 = ["正常", "侧视", "喝水", "吸烟", "操作中控", "玩手机", "侧身拿东西", "接电话"]
labels_9 = ["正常", "侧视", "喝水", "吸烟", "操作中控", "玩手机", "侧身拿东西", "整理仪容", "接电话"]
video_path = r"data/video/xw.mp4"
def softmax_np(x):
x_row_max = x.max(axis=-1)
x_row_max = x_row_max.reshape(list(x.shape)[:-1]+[1])
x = x - x_row_max
x_exp = np.exp(x)
x_exp_row_sum = x_exp.sum(axis=-1).reshape(list(x.shape)[:-1]+[1])
softmax = x_exp / x_exp_row_sum
return softmax
def detect_img():
font = ImageFont.truetype(font='data/video/simhei.ttf', size=int(30)) # 20
while True:
img = input('Input image filename:')
try:
frame = cv2.imread(img)
frame2 = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
except:
print('Open Error! Try again!')
continue
else:
# crop crop_bbox = (0, 0, 1252, 1296) 2304*1296
# ~ cw = int(1252/2304*640)
# ~ ch = 360
# ~ frame2 = frame2[0:ch,0:cw]
start = time.time()
image = cv2.resize(frame2, (224, 224))
# image = cv2.resize(frame2, (160, 160))
# print("resize time:", time.time()-start)
image = np.float32(image) / 255.0
image[:, :, ] -= (np.float32(0.485), np.float32(0.456), np.float32(0.406))
image[:, :, ] /= (np.float32(0.229), np.float32(0.224), np.float32(0.225))
image = image.transpose(2, 0, 1) # 转换轴，pytorch为channel first
image = image.reshape(1, 3, 224, 224) # barch,channel,height,weight
# image = image.reshape(1, 3, 160, 160) # barch,channel,height,weight
inputs = {onnx_session.get_inputs()[0].name: image}
print("preprocess time:", time.time() - start)
# 注意返回为三维数组（1,1,class_num)
probs = onnx_session.run(None, inputs)
index = np.argmax(probs)
# print(probs)
softmax_probs = softmax_np(np.array(probs))
prob = softmax_probs.max()
# print(index,prob)
# index =0
print(labels_9[index], prob)
predict_time = time.time() - start
print("preprocess time:", predict_time)
img_PIL = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_PIL)
text = "{}:{:.2f} 检测时间：{:.3f}s".format(labels_9[index], prob, predict_time)
draw.text((100, 30), text, fill=(255, 0, 0), font=font)
# img_PIL.show()
frame = cv2.cvtColor(np.asarray(img_PIL), cv2.COLOR_RGB2BGR)
cv2.imshow("img", frame)
cv2.waitKey(0)
# cv2.imshow("result", frame)
# if cv2.waitKey(0) & 0xFF == ord('q'):
# break
def detect_video(output_path=""):
vid = cv2.VideoCapture(video_path)
if not vid.isOpened():
raise IOError("Couldn't open webcam or video")
video_FourCC = int(vid.get(cv2.CAP_PROP_FOURCC))
video_fps = vid.get(cv2.CAP_PROP_FPS)
video_size = (int(vid.get(cv2.CAP_PROP_FRAME_WIDTH)),
int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)))
isOutput = True if output_path != "" else False
if isOutput:
print("!!! TYPE:", type(output_path), type(
video_FourCC), type(video_fps), type(video_size))
print("!!! TYPE:", output_path,
video_FourCC, video_fps, video_size)
out = cv2.VideoWriter(output_path, video_FourCC, video_fps, video_size)
accum_time = 0
detect_fps = 0
d_fps = "detect_FPS: ??"
prev_time = time.time()
font = ImageFont.truetype(font='data/video/simhei.ttf', size=int(30)) # 20
while True:
return_value, frame = vid.read()
if not return_value:
print("Can't receive frame (stream end?). Exiting ...")
break
start = time.time()
frame2 = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# crop crop_bbox = (0, 0, 1252, 1296) 2304*1296
# ~ cw = int(1252/2304*640)
# ~ ch = 360
# ~ frame2 = frame2[0:ch,0:cw]
image = cv2.resize(frame2, (224, 224))
# image = cv2.resize(frame2, (160, 160))
# print("resize time:", time.time()-start)
image = np.float32(image) / 255.0
image[:, :, ] -= (np.float32(0.485), np.float32(0.456), np.float32(0.406))
image[:, :, ] /= (np.float32(0.229), np.float32(0.224), np.float32(0.225))
image = image.transpose(2, 0, 1) # 转换轴，pytorch为channel first
image = image.reshape(1, 3, 224, 224) # barch,channel,height,weight
# image = image.reshape(1, 3, 160, 160) # barch,channel,height,weight
inputs = {onnx_session.get_inputs()[0].name: image}
print("preprocess time:", time.time() -start)
# 注意返回为三维数组（1,1,class_num)
probs = onnx_session.run(None, inputs)
print("predict time:", time.time() - start)
index = np.argmax(probs)
# print(probs)
softmax_probs = softmax_np(np.array(probs))
prob = softmax_probs.max()
# print(index,prob)
# index =0
print(labels_9[index], prob)
img_PIL = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_PIL)
detect_fps = detect_fps + 1
curr_time = time.time()
exec_time = curr_time - prev_time
prev_time = curr_time
accum_time = accum_time + exec_time
if accum_time > 1:
accum_time = accum_time - 1
d_fps = "detect_FPS:" + str(detect_fps)
detect_fps = 0
text = "{}:{:.2f} {}".format(labels_9[index], prob, d_fps)
draw.text((200, 30), text, fill=(255, 0, 0), font=font)
frame = cv2.cvtColor(np.asarray(img_PIL), cv2.COLOR_RGB2BGR)
cv2.namedWindow("result", cv2.WINDOW_NORMAL)
cv2.imshow("result", frame)
if isOutput:
out.write(frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
if __name__ == '__main__':
# detect_video(r"data/video/test.mp4")
# detect_video()
detect_img()