用堆叠自编码器对文件夹内的图像进行分类Python代码

使用堆叠自编码器对文件夹内的图像进行分类的Python代码如下：

import os
import numpy as np
from sklearn.model_selection import train_test_split
from keras.layers import Input, Dense
from keras.models import Model
from keras.optimizers import Adam
from keras.utils import to_categorical
from keras.preprocessing.image import load_img, img_to_array

# 设置文件夹路径和类别
folder_path = 'path/to/folder'
categories = ['category1', 'category2', 'category3']

# 加载图像并按类别标记
X = []
y = []
for category in categories:
    for filename in os.listdir(os.path.join(folder_path, category)):
        img = load_img(os.path.join(folder_path, category, filename), target_size=(224, 224))
        img_arr = img_to_array(img)
        X.append(img_arr)
        y.append(categories.index(category))

# 将数据分为训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# 将类别标记转换为独热编码
num_classes = len(categories)
y_train = to_categorical(y_train, num_classes)
y_test = to_categorical(y_test, num_classes)

# 定义堆叠自编码器
input_img = Input(shape=(224, 224, 3))
x = Dense(128, activation='relu')(input_img)
x = Dense(64, activation='relu')(x)
encoded = Dense(32, activation='relu')(x)
x = Dense(64, activation='relu')(encoded)
x = Dense(128, activation='relu')(x)
decoded = Dense(3, activation='sigmoid')(x)
autoencoder = Model(input_img, decoded)

# 编译自编码器
autoencoder.compile(optimizer=Adam(lr=0.001), loss='mse')

# 训练自编码器
autoencoder.fit(X_train, X_train, epochs=50, batch_size=32, validation_data=(X_test, X_test))

# 提取编码层和分类器
encoder = Model(input_img, encoded)
x = Dense(256, activation='relu')(encoded)
x = Dense(128, activation='relu')(x)
x = Dense(64, activation='relu')(x)
output = Dense(num_classes, activation='softmax')(x)
classifier = Model(input_img, output)

# 编译分类器
classifier.compile(optimizer=Adam(lr=0.001), loss='categorical_crossentropy', metrics=['accuracy'])

# 训练分类器
classifier.fit(X_train, y_train, epochs=50, batch_size=32, validation_data=(X_test, y_test))

此代码使用Keras库构建堆叠自编码器和分类器模型，对文件夹内的图像进行分类。首先，将图像加载到列表中，并按类别标记。然后将数据分为训练集和测试集，并将类别标记转换为独热编码。接下来，定义一个堆叠自编码器模型和一个编码器模型来提取编码层特征。使用训练集对自编码器进行训练，并定期评估重构误差。然后使用编码器提取特征，并将其馈入一个分类器模型，该模型包含几个密集层和一个softmax激活输出层，用于对图像进行分类。最后，使用训练集对分类器进行训练，并在测试集上进行评估。