caffe/gpu_mode.py at master · 541435721/caffe · GitHub

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
# coding:utf-8
# __author__ = 'BianXuesheng'
# __data__ = '2016_2016/2/28_15:30 '

import numpy as np
import matplotlib.pyplot as plt
import os
import sys
import caffe
import time
import cv2

caffe_root = 'E:/caffe-windows-master/'

plt.rcParams['figure.figsize'] = (10, 10)
plt.rcParams['image.interpolation'] = 'nearest'
plt.rcParams['image.cmap'] = 'gray'

start = time.clock()

# 设置GPU模式
caffe.set_device(0)
caffe.set_mode_gpu()
# 用模型创建初始化网络
net = caffe.Net(caffe_root + 'models/bvlc_reference_caffenet/deploy.prototxt',
                caffe_root + 'models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel',
                caffe.TEST)

# 输入预处理 'data'是输入blob对象的名字，blobs['data']==net.input[0]
transformer = caffe.io.Transformer({'data': net.blobs['data'].data.shape})
transformer.set_transpose('data', (2, 0, 1))
transformer.set_mean('data',
                     np.load(caffe_root + 'python/caffe/imagenet/ilsvrc_2012_mean.npy').mean(1).mean(1))  # mean pixel
transformer.set_raw_scale('data', 255)  # 参考模型作用于像素值为[0,255]的范围，而不是[0,1]
transformer.set_channel_swap('data', (2, 1, 0))  # 参考模型作用的像素通道顺序为BGR而不是RGB

# 设置bacth大小为50
net.blobs['data'].reshape(50, 3, 227, 227)

# 读入将要进行分类的图片
net.blobs['data'].data[...] = transformer.preprocess('data',
                                                     caffe.io.load_image(caffe_root + 'examples/images/cat.jpg'))
# 图像往上层前进
out = net.forward()
print("Predicted class is #{}.".format(out['prob'][0].argmax()))

# 显示图片
cv2.imshow('test', transformer.deprocess('data', net.blobs['data'].data[0]))
cv2.waitKey()

# 加载分类标签
imagenet_labels_filename = caffe_root + 'data/ilsvrc12/synset_words.txt'
labels = np.loadtxt(imagenet_labels_filename, str, delimiter='\t')

top_k = net.blobs['prob'].data[0].flatten().argsort()[-1:-6:-1]
print labels[top_k]

# print time.clock() - start

net.forward()

# 打印出每层输出名称和形状
print [(k, v.data.shape) for k, v in net.blobs.items()]


# 可视化函数；获取每层输出的array.
def vis_square(data, padsize=1, padval=0):
    data -= data.min()
    data /= data.max()

    # force the number of filters to be square
    n = int(np.ceil(np.sqrt(data.shape[0])))  # 获取图像的宽度尺寸
    padding = ((0, n ** 2 - data.shape[0]), (0, padsize), (0, padsize)) + ((0, 0),) * (data.ndim - 3)
    data = np.pad(data, padding, mode='constant', constant_values=(padval, padval))

    # tile the filters into an image
    data = data.reshape((n, n) + data.shape[1:]).transpose((0, 2, 1, 3) + tuple(range(4, data.ndim + 1)))
    data = data.reshape((n * data.shape[1], n * data.shape[3]) + data.shape[4:])

    # plt.imshow(data)
    cv2.imshow('test', data)
    cv2.waitKey(3000)


# 卷积核，[weights,biases]构成的list
filters = net.params['conv1'][0].data
vis_square(filters.transpose(0, 2, 3, 1))

# 卷积后的输出图像，先输出前36张
feat = net.blobs['conv1'].data[0, :36]
vis_square(feat, padval=1)

# 第二个卷积核,一共256个
filters = net.params['conv2'][0].data
vis_square(filters[:48].reshape(48 ** 2, 5, 5))

# 卷积核2的输出结果 共256个
feat = net.blobs['conv2'].data[0, :36]
vis_square(feat, padval=1)

# 卷积核3输出结果
feat = net.blobs['conv3'].data[0]
vis_square(feat, padval=0.5)

# 卷积核4输出结果
feat = net.blobs['conv4'].data[0]
vis_square(feat, padval=0.5)

# 卷积核5输出结果
feat = net.blobs['conv5'].data[0]
vis_square(feat, padval=0.5)

# 池化层5输出结果
feat = net.blobs['pool5'].data[0]
vis_square(feat, padval=1)

# 第一层全连接层输出结果
feat = net.blobs['fc6'].data[0]
plt.subplot(2, 1, 1)
plt.plot(feat.flat)
plt.subplot(2, 1, 2)
_ = plt.hist(feat.flat[feat.flat > 0], bins=100)
plt.show()

# 第二层全连接层输出结果
feat = net.blobs['fc7'].data[0]
plt.subplot(2, 1, 1)
plt.plot(feat.flat)
plt.subplot(2, 1, 2)
_ = plt.hist(feat.flat[feat.flat > 0], bins=100)
plt.show()

# 最终结果输出
feat = net.blobs['prob'].data[0]
plt.plot(feat.flat)
plt.show()