TensorFlowCNN中间层可视化

import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt


fashion_mnist = keras.datasets.fashion_mnist
(x_train,y_train),(x_test,y_test) = fashion_mnist.load_data()
model2 = keras.Sequential()
model2.add(keras.layers.Conv2D(64,(3,3),activation='relu',input_shape=(28,28,1)))
model2.add(keras.layers.MaxPool2D(2,2))
model2.add(keras.layers.Conv2D(64,(3,3),activation='relu'))
model2.add(keras.layers.MaxPool2D(2,2))
model2.add(keras.layers.Flatten())
model2.add(keras.layers.Dense(128,activation=tf.nn.relu))
model2.add(keras.layers.Dense(10,activation=tf.nn.softmax))
model2.compile(optimizer=tf.optimizers.Adam(),loss=tf.losses.sparse_categorical_crossentropy,
               metrics=['accuracy'])
model2.fit(x_train,y_train,epochs=10,verbose=2)
model2.summary()

# 可视化
layer_outs = [layer.output for layer in model2.layers]
activation_model = tf.keras.models.Model(inputs=model2.input,outputs = layer_outs)
pred = activation_model.predict(x_train[0].reshape(1,28,28,1))
len(pred)
# pred中包含每个层的输出
pred[0].shape
plt.imshow(pred[0][0,:,:,0])