图片的自动编码很容易就想到用卷积神经网络做为编码-解码器。在实际的操作中，
也经常使用卷积自动编码器去解决图像编码问题，而且非常有效。
下面通过**keras**完成简单的卷积自动编码。 编码器有堆叠的卷积层和池化层(max pooling用于空间降采样)组成。 对应的解码器由卷积层和上采样层组成。
|
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
|
@requires_authorization# -*- coding:utf-8 -*-from keras.layers import Input, Dense, Conv2D, MaxPooling2D, UpSampling2Dfrom keras.models import Modelfrom keras import backend as Kimport os## 网络结构 ##input_img = Input(shape=(28,28,1)) # Tensorflow后端， 注意要用channel_last# 编码器部分x = Conv2D(16, (3,3), activation='relu', padding='same')(input_img)x = MaxPooling2D((2,2), padding='same')(x)x = Conv2D(8,(3,3), activation='relu', padding='same')(x)x = MaxPooling2D((2,2), padding='same')(x)x = Conv2D(8, (3,3), activation='relu', padding='same')(x)encoded = MaxPooling2D((2,2), padding='same')(x)# 解码器部分x = Conv2D(8, (3,3), activation='relu', padding='same')(encoded)x = UpSampling2D((2, 2))(x)x = Conv2D(8, (3,3), activation='relu', padding='same')(x) x = UpSampling2D((2, 2))(x)x = Conv2D(16, (3, 3), activation='relu'， padding='same')(x)x = UpSampling2D((2, 2))(x)decoded = Conv2D(1, (3, 3), activation='sigmoid', padding='same')(x)autoencoder = Model(input_img, decoded)autoencoder.compile(optimizer='adam', loss='binary_crossentropy')# 得到编码层的输出encoder_model = Model(inputs=autoencoder.input, outputs=autoencoder.get_layer('encoder_out').output)## 导入数据， 使用常用的手写识别数据集def load_mnist(dataset_name):'''load the data''' data_dir = os.path.join("./data", dataset_name) f = np.load(os.path.join(data_dir, 'mnist.npz')) train_data = f['train'].T trX = train_data.reshape((-1, 28, 28, 1)).astype(np.float32) trY = f['train_labels'][-1].astype(np.float32) test_data = f['test'].T teX = test_data.reshape((-1, 28, 28, 1)).astype(np.float32) teY = f['test_labels'][-1].astype(np.float32) # one-hot # y_vec = np.zeros((len(y), 10), dtype=np.float32) # for i, label in enumerate(y): # y_vec[i, y[i]] = 1 # keras.utils里带的有one-hot的函数， 就直接用那个了 return trX / 255., trY, teX/255., teY# 开始导入数据x_train, _ , x_test, _= load_mnist('mnist')# 可视化训练结果， 我们打开终端， 使用tensorboard# tensorboard --logdir=/tmp/autoencoder # 注意这里是打开一个终端， 在终端里运行# 训练模型， 并且在callbacks中使用tensorBoard实例， 写入训练日志 http://0.0.0.0:6006from keras.callbacks import TensorBoardautoencoder.fit(x_train, x_train, epochs=50, batch_size=128, shuffle=True, validation_data=(x_test, x_test), callbacks=[TensorBoard(log_dir='/tmp/autoencoder')])# 重建图片import matplotlib.pyplot as plt decoded_imgs = autoencoder.predict(x_test)encoded_imgs = encoder_model.predict(x_test)n = 10plt.figure(figsize=(20, 4))for i in range(n): k = i + 1 # 画原始图片 ax = plt.subplot(2, n, k) plt.imshow(x_test[k].reshape(28, 28)) plt.gray() ax.get_xaxis().set_visible(False) # 画重建图片 ax = plt.subplot(2, n, k + n) plt.imshow(decoded_imgs[i].reshape(28, 28)) plt.gray() ax.get_xaxis().set_visible(False) ax.get_yaxis().set_visible(False)plt.show()# 编码得到的特征n = 10plt.figure(figsize=(20, 8))for i in range(n): k = i + 1 ax = plt.subplot(1, n, k) plt.imshow(encoded[k].reshape(4, 4 * 8).T) plt.gray() ax.get_xaxis().set_visible(False) ax.get_yaxis().set_visible(False)plt.show() |
补充知识：keras搬砖系列-单层卷积自编码器
考试成绩出来了，竟然有一门出奇的差，只是有点意外。
觉得应该不错的，竟然考差了，它估计写了个随机数吧。
头文件
|
1
2
3
4
5
|
from keras.layers import Input,Densefrom keras.models import Model from keras.datasets import mnistimport numpy as np import matplotlib.pyplot as plt |
导入数据
|
1
2
3
4
5
6
|
(X_train,_),(X_test,_) = mnist.load_data() X_train = X_train.astype('float32')/255.X_test = X_test.astype('float32')/255.X_train = X_train.reshape((len(X_train),-1))X_test = X_test.reshape((len(X_test),-1)) |
这里的X_train和X_test的维度分别为(60000L,784L),(10000L,784L)
这里进行了归一化，将所有的数值除上255.
设定编码的维数与输入数据的维数
encoding_dim = 32
input_img = Input(shape=(784,))
构建模型
|
1
2
3
4
5
6
7
8
9
|
encoded = Dense(encoding_dim,activation='relu')(input_img)decoded = Dense(784,activation='relu')(encoded) autoencoder = Model(inputs = input_img,outputs=decoded)encoder = Model(inputs=input_img,outputs=encoded) encoded_input = Input(shape=(encoding_dim,))decoder_layer = autoencoder.layers[-1]deconder = Model(inputs=encoded_input,outputs = decoder_layer(encoded_input)) |
模型编译
autoencoder.compile(optimizer='adadelta',loss='binary_crossentropy')
模型训练
autoencoder.fit(X_train,X_train,epochs=50,batch_size=256,shuffle=True,validation_data=(X_test,X_test))
预测
encoded_imgs = encoder.predict(X_test)
decoded_imgs = deconder.predict(encoded_imgs)
数据可视化
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
|
n = 10for i in range(n): ax = plt.subplot(2,n,i+1) plt.imshow(X_test[i].reshape(28,28)) plt.gray() ax.get_xaxis().set_visible(False) ax.get_yaxis().set_visible(False) ax = plt.subplot(2,n,i+1+n) plt.imshow(decoded_imgs[i].reshape(28,28)) plt.gray() ax.get_xaxis().set_visible(False) ax.get_yaxis().set_visible(False)plt.show() |
完成代码
|
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
|
from keras.layers import Input,Densefrom keras.models import Model from keras.datasets import mnistimport numpy as np import matplotlib.pyplot as plt (X_train,_),(X_test,_) = mnist.load_data() X_train = X_train.astype('float32')/255.X_test = X_test.astype('float32')/255.X_train = X_train.reshape((len(X_train),-1))X_test = X_test.reshape((len(X_test),-1)) encoding_dim = 32input_img = Input(shape=(784,)) encoded = Dense(encoding_dim,activation='relu')(input_img)decoded = Dense(784,activation='relu')(encoded) autoencoder = Model(inputs = input_img,outputs=decoded)encoder = Model(inputs=input_img,outputs=encoded) encoded_input = Input(shape=(encoding_dim,))decoder_layer = autoencoder.layers[-1]deconder = Model(inputs=encoded_input,outputs = decoder_layer(encoded_input)) autoencoder.compile(optimizer='adadelta',loss='binary_crossentropy')autoencoder.fit(X_train,X_train,epochs=50,batch_size=256,shuffle=True,validation_data=(X_test,X_test)) encoded_imgs = encoder.predict(X_test)decoded_imgs = deconder.predict(encoded_imgs) ##vian = 10for i in range(n): ax = plt.subplot(2,n,i+1) plt.imshow(X_test[i].reshape(28,28)) plt.gray() ax.get_xaxis().set_visible(False) ax.get_yaxis().set_visible(False) ax = plt.subplot(2,n,i+1+n) plt.imshow(decoded_imgs[i].reshape(28,28)) plt.gray() ax.get_xaxis().set_visible(False) ax.get_yaxis().set_visible(False)plt.show() |
以上这篇keras自动编码器实现系列之卷积自动编码器操作就是小编分享给大家的全部内容了，希望能给大家一个参考，也希望大家多多支持服务器之家。
原文链接：https://blog.csdn.net/xuyang508/article/details/74276181
