Pytorch: 利用预训练的残差网络ResNet50进行图像特征提取，并可视化特征图&热图

1. 残差网络ResNet的结构

2.图像特征提取和可视化分析

import cv2
import time
import os
import matplotlib.pyplot as plt
import torch
from torch import nn
import torchvision.models as models
import torchvision.transforms as transforms
import numpy as np

imgname = 'bottle_broken_large.png' 
savepath='vis_resnet50/features_bottle'
if not os.path.isdir(savepath):
    os.makedirs(savepath)

def draw_features(width,height,x,savename):
    tic = time.time()
    fig = plt.figure(figsize=(16, 16))
    fig.subplots_adjust(left=0.05, right=0.95, bottom=0.05, top=0.95, wspace=0.05, hspace=0.05)
    for i in range(width*height):
        plt.subplot(height, width, i + 1)
        plt.axis('off')
        img = x[0, i, :, :]
        pmin = np.min(img)
        pmax = np.max(img)
        img = ((img - pmin) / (pmax - pmin + 0.000001))*255  #float在[0，1]之间，转换成0-255
        img=img.astype(np.uint8)  #转成unit8
        img=cv2.applyColorMap(img, cv2.COLORMAP_JET) #生成heat map
        img = img[:, :, ::-1]#注意cv2（BGR）和matplotlib(RGB)通道是相反的
        plt.imshow(img)
        print("{}/{}".format(i,width*height))
    fig.savefig(savename, dpi=100)
    fig.clf()
    plt.close()
    print("time:{}".format(time.time()-tic))


class ft_net(nn.Module):

    def __init__(self):
        super(ft_net, self).__init__()
        model_ft = models.resnet50(pretrained=True)
        self.model = model_ft

    def forward(self, x):
        if True: # draw features or not
            x = self.model.conv1(x)
            draw_features(8, 8, x.cpu().numpy(),"{}/f1_conv1.png".format(savepath))

            x = self.model.bn1(x)
            draw_features(8, 8, x.cpu().numpy(),"{}/f2_bn1.png".format(savepath))

            x = self.model.relu(x)
            draw_features(8, 8, x.cpu().numpy(), "{}/f3_relu.png".format(savepath))

            x = self.model.maxpool(x)
            draw_features(8, 8, x.cpu().numpy(), "{}/f4_maxpool.png".format(savepath))

            x = self.model.layer1(x)
            draw_features(16, 16, x.cpu().numpy(), "{}/f5_layer1.png".format(savepath))

            x = self.model.layer2(x)
            draw_features(16, 32, x.cpu().numpy(), "{}/f6_layer2.png".format(savepath))

            x = self.model.layer3(x)
            draw_features(32, 32, x.cpu().numpy(), "{}/f7_layer3.png".format(savepath))

            x = self.model.layer4(x)
            draw_features(32, 32, x.cpu().numpy()[:, 0:1024, :, :], "{}/f8_layer4_1.png".format(savepath))
            draw_features(32, 32, x.cpu().numpy()[:, 1024:2048, :, :], "{}/f8_layer4_2.png".format(savepath))

            x = self.model.avgpool(x)
            plt.plot(np.linspace(1, 2048, 2048), x.cpu().numpy()[0, :, 0, 0])
            plt.savefig("{}/f9_avgpool.png".format(savepath))
            plt.clf()
            plt.close()

            x = x.view(x.size(0), -1)
            x = self.model.fc(x)
            plt.plot(np.linspace(1, 1000, 1000), x.cpu().numpy()[0, :])
            plt.savefig("{}/f10_fc.png".format(savepath))
            plt.clf()
            plt.close()
        else :
            x = self.model.conv1(x)
            x = self.model.bn1(x)
            x = self.model.relu(x)
            x = self.model.maxpool(x)
            x = self.model.layer1(x)
            x = self.model.layer2(x)
            x = self.model.layer3(x)
            x = self.model.layer4(x)
            x = self.model.avgpool(x)
            x = x.view(x.size(0), -1)
            x = self.model.fc(x)

        return x


model = ft_net().cuda()

# pretrained_dict = resnet50.state_dict()
# pretrained_dict = {k: v for k, v in pretrained_dict.items() if k in model_dict}
# model_dict.update(pretrained_dict)
# net.load_state_dict(model_dict)
model.eval()
img = cv2.imread(imgname)
img = cv2.resize(img, (288, 288))
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
transform = transforms.Compose(
    [transforms.ToTensor(),
     transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
img = transform(img).cuda()
img = img.unsqueeze(0)

with torch.no_grad():
    start = time.time()
    out = model(img)
    print("total time:{}".format(time.time()-start))
    result = out.cpu().numpy()
    # ind=np.argmax(out.cpu().numpy())
    ind = np.argsort(result, axis=1)
    for i in range(5):
        print("predict:top {} = cls {} : score {}".format(i+1,ind[0,1000-i-1],result[0,1000-i-1]))
    print("done")
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

可视化结果：