-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathpredict.py
149 lines (129 loc) · 4.68 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
'''
**************************************************
@File ：AttitudeRecognition -> predict
@IDE ：PyCharm
@Author ：TheOnlyMan
@Date ：2023/4/20 10:54
**************************************************
'''
import os
import random
import sys
import time
from math import ceil
import cv2
import matplotlib.pyplot as plt
import numpy as np
import torch
from arguments import init
from cpn.network import cpn
from utils.image_utils import load_image, im_to_torch, to_numpy, display_heatmap
from utils.model_utils import switch
dirname = os.path.dirname(__file__)
sys.path.append(os.path.abspath(os.path.join(dirname, 'cpn')))
seed = 3407
torch.manual_seed(seed)
np.random.seed(seed)
random.seed(seed)
torch.cuda.manual_seed_all(seed)
def get_keypoints(refine_output, flip_refine_output=None):
single_map = to_numpy(refine_output)
if flip_refine_output is not None:
flip_single_map = to_numpy(flip_refine_output)
fscore = flip_single_map.transpose((1, 2, 0))
fscore = cv2.flip(fscore, 1)
fscore = fscore.transpose((2, 0, 1))
fscore = switch(fscore)
single_map += fscore
single_map /= 2
r0 = single_map.copy()
r0 /= (r0.max() - r0.min())
v_score = np.zeros(17)
keypoints = []
heatmap = single_map.copy()
for p in range(17):
single_map[p] /= np.amax(single_map[p])
heatmap[p] /= np.amax(heatmap[p])
border = 10
dr = np.zeros((64 + 2 * border, 48 + 2 * border))
dr[border:-border, border:-border] = single_map[p].copy()
dr = cv2.GaussianBlur(dr, (21, 21), 0)
lb = dr.argmax()
y, x = np.unravel_index(lb, dr.shape)
dr[y, x] = 0
lb = dr.argmax()
py, px = np.unravel_index(lb, dr.shape)
y -= border
x -= border
py -= border + y
px -= border + x
ln = (px ** 2 + py ** 2) ** 0.5
delta = 0.25
if ln > 1e-3:
x += delta * px / ln
y += delta * py / ln
x = max(0, min(x, 48 - 1))
y = max(0, min(y, 64 - 1))
resx = float((4 * x + 2) / 192)
resy = float((4 * y + 2) / 256)
v_score[p] = float(r0[p, int(round(y) + 1e-10), int(round(x) + 1e-10)])
keypoints.append([resx, resy, v_score[p]])
return np.array(keypoints), heatmap
def pred(args):
model_path, image_path = args.model, args.image
model = cpn((64, 48), args).to(args.device)
model.load_state_dict(torch.load(model_path))
model.eval()
image = load_image(image_path)
new_image = cv2.resize(image, (192, 256), cv2.INTER_LANCZOS4)
input = im_to_torch(new_image)
pixel_means = (np.array([102.9801, 115.9465, 122.7717], dtype=np.float32) / 255.0).reshape(3, 1, 1) # BGR
input -= pixel_means
input = input.view(1, *input.shape).to(args.device)
ratio = image.shape[0] / 256
radius = ceil(5 * ratio)
thickness = ceil(2 * ratio)
global_output, refine_output = model(input)
if not args.noflip:
flip_image = new_image.copy()
flip_image = cv2.flip(flip_image, 1)
input_flip = im_to_torch(flip_image)
input_flip -= pixel_means
input_flip = input_flip.view(1, *input_flip.shape).to(args.device)
flip_global_output, flip_refine_output = model(input_flip)
keypoints, heatmap = get_keypoints(refine_output[0], flip_refine_output[0])
else:
keypoints, heatmap = get_keypoints(refine_output[0])
image = np.array(image)
colors = [
(255, 0, 0), # 蓝色
(0, 255, 0), # 绿色
(0, 0, 255), # 红色
(255, 255, 0), # 黄色
(255, 0, 255), # 粉色
(0, 255, 255), # 青色
(255, 165, 0), # 橙色
(128, 0, 128), # 紫色
(240, 248, 255) # 淡蓝色
]
skeleton = [[16, 14], [14, 12], [17, 15], [15, 13], [12, 13], [6, 12], [7, 13], [6, 7], [6, 8], [7, 9],
[8, 10], [9, 11], [2, 3], [1, 2], [1, 3], [2, 4], [3, 5], [4, 6], [5, 7]]
for i in range(17):
x, y, v = keypoints[i]
if v > args.lim:
cv2.circle(image, (int(x * image.shape[1]), int(y * image.shape[0])), radius, colors[(i + 1) // 2], thickness)
for i, j in skeleton:
x1, y1, v1 = keypoints[i - 1]
x2, y2, v2 = keypoints[j - 1]
if v1 > args.lim and v2 > args.lim:
x1, y1 = int(x1 * image.shape[1]), int(y1 * image.shape[0])
x2, y2 = int(x2 * image.shape[1]), int(y2 * image.shape[0])
cv2.line(image, (x1, y1), (x2, y2), colors[1], thickness)
plt.imshow(image[:, :, ::-1])
plt.axis('off')
plt.show()
print(heatmap.shape)
display_heatmap(heatmap, image, keypoints)
if __name__ == '__main__':
args = init('predict')
pred(args)