参考1：【推荐】怎么显示Keras的网络结构和其中的参数
参考2：【推荐】Mac BigSur：安装homebrew(国内源)+Graphviz
也可以在线实现，https://netron.app，需要 *.h5 文件，信息更智能，但是没有输入，不过用了不同颜色显示。（示例图在文末）
1. 安装 pydot
pip install pydot
2. 安装 pydot-ng
pip install pydot-ng
3. 安装 homebrew（国内源）
打开终端，输入以下代码：
/bin/zsh -c "$(curl -fsSL https://gitee.com/cunkai/HomebrewCN/raw/master/Homebrew.sh)"
选择序号：1
执行脚本：Y
4. 安装 Graphviz
homebrew安装完毕后，运行 brew install graphviz即可
brew install graphviz
参考2可以有效解决
举例：
from tensorflow.keras import layers
def get_model(img_size, num_classes):
# 二维变三维，元组加法相当于 concat
inputs = keras.Input(shape=img_size + (3,))
### [First half of the network: downsampling inputs] ###
# Entry block
x = layers.Conv2D(32, 3, strides=2, padding="same")(inputs)
x = layers.BatchNormalization()(x)
x = layers.Activation("relu")(x)
previous_block_activation = x # Set aside residual
# Blocks 1, 2, 3 are identical apart from the feature depth.
for filters in [64, 128, 256]:
x = layers.Activation("relu")(x)
x = layers.SeparableConv2D(filters, 3, padding="same")(x)
x = layers.BatchNormalization()(x)
x = layers.Activation("relu")(x)
x = layers.SeparableConv2D(filters, 3, padding="same")(x)
x = layers.BatchNormalization()(x)
x = layers.MaxPooling2D(3, strides=2, padding="same")(x)
# Project residual
residual = layers.Conv2D(filters, 1, strides=2, padding="same")(
previous_block_activation
)
x = layers.add([x, residual]) # Add back residual
previous_block_activation = x # Set aside next residual
### [Second half of the network: upsampling inputs] ###
for filters in [256, 128, 64, 32]:
x = layers.Activation("relu")(x)
x = layers.Conv2DTranspose(filters, 3, padding="same")(x)
x = layers.BatchNormalization()(x)
x = layers.Activation("relu")(x)
x = layers.Conv2DTranspose(filters, 3, padding="same")(x)
x = layers.BatchNormalization()(x)
x = layers.UpSampling2D(2)(x)
# Project residual
residual = layers.UpSampling2D(2)(previous_block_activation)
residual = layers.Conv2D(filters, 1, padding="same")(residual)
x = layers.add([x, residual]) # Add back residual
previous_block_activation = x # Set aside next residual
# Add a per-pixel classification layer
outputs = layers.Conv2D(num_classes, 3, activation="softmax", padding="same")(x)
# Define the model
model = keras.Model(inputs, outputs)
return model
# Free up RAM in case the model definition cells were run multiple times
keras.backend.clear_session()
# Build model
model = get_model(img_size, num_classes)
model.summary()
# 结构图显示
from keras.utils.vis_utils import plot_model
plot_model(model, to_file='Flatten.png', show_shapes=True)
输出
Model: "model"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) [(None, 160, 160, 3) 0
__________________________________________________________________________________________________
conv2d (Conv2D) (None, 80, 80, 32) 896 input_1[0][0]
__________________________________________________________________________________________________
batch_normalization (BatchNorma (None, 80, 80, 32) 128 conv2d[0][0]
__________________________________________________________________________________________________
activation (Activation) (None, 80, 80, 32) 0 batch_normalization[0][0]
__________________________________________________________________________________________________
activation_1 (Activation) (None, 80, 80, 32) 0 activation[0][0]
__________________________________________________________________________________________________
separable_conv2d (SeparableConv (None, 80, 80, 64) 2400 activation_1[0][0]
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 80, 80, 64) 256 separable_conv2d[0][0]
__________________________________________________________________________________________________
activation_2 (Activation) (None, 80, 80, 64) 0 batch_normalization_1[0][0]
__________________________________________________________________________________________________
separable_conv2d_1 (SeparableCo (None, 80, 80, 64) 4736 activation_2[0][0]
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 80, 80, 64) 256 separable_conv2d_1[0][0]
__________________________________________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 40, 40, 64) 0 batch_normalization_2[0][0]
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 40, 40, 64) 2112 activation[0][0]
__________________________________________________________________________________________________
add (Add) (None, 40, 40, 64) 0 max_pooling2d[0][0]
conv2d_1[0][0]
__________________________________________________________________________________________________
activation_3 (Activation) (None, 40, 40, 64) 0 add[0][0]
__________________________________________________________________________________________________
separable_conv2d_2 (SeparableCo (None, 40, 40, 128) 8896 activation_3[0][0]
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 40, 40, 128) 512 separable_conv2d_2[0][0]
__________________________________________________________________________________________________
activation_4 (Activation) (None, 40, 40, 128) 0 batch_normalization_3[0][0]
__________________________________________________________________________________________________
separable_conv2d_3 (SeparableCo (None, 40, 40, 128) 17664 activation_4[0][0]
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 40, 40, 128) 512 separable_conv2d_3[0][0]
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 20, 20, 128) 0 batch_normalization_4[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 20, 20, 128) 8320 add[0][0]
__________________________________________________________________________________________________
add_1 (Add) (None, 20, 20, 128) 0 max_pooling2d_1[0][0]
conv2d_2[0][0]
__________________________________________________________________________________________________
activation_5 (Activation) (None, 20, 20, 128) 0 add_1[0][0]
__________________________________________________________________________________________________
separable_conv2d_4 (SeparableCo (None, 20, 20, 256) 34176 activation_5[0][0]
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 20, 20, 256) 1024 separable_conv2d_4[0][0]
__________________________________________________________________________________________________
activation_6 (Activation) (None, 20, 20, 256) 0 batch_normalization_5[0][0]
__________________________________________________________________________________________________
separable_conv2d_5 (SeparableCo (None, 20, 20, 256) 68096 activation_6[0][0]
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 20, 20, 256) 1024 separable_conv2d_5[0][0]
__________________________________________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 10, 10, 256) 0 batch_normalization_6[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 10, 10, 256) 33024 add_1[0][0]
__________________________________________________________________________________________________
add_2 (Add) (None, 10, 10, 256) 0 max_pooling2d_2[0][0]
conv2d_3[0][0]
__________________________________________________________________________________________________
activation_7 (Activation) (None, 10, 10, 256) 0 add_2[0][0]
__________________________________________________________________________________________________
conv2d_transpose (Conv2DTranspo (None, 10, 10, 256) 590080 activation_7[0][0]
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 10, 10, 256) 1024 conv2d_transpose[0][0]
__________________________________________________________________________________________________
activation_8 (Activation) (None, 10, 10, 256) 0 batch_normalization_7[0][0]
__________________________________________________________________________________________________
conv2d_transpose_1 (Conv2DTrans (None, 10, 10, 256) 590080 activation_8[0][0]
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 10, 10, 256) 1024 conv2d_transpose_1[0][0]
__________________________________________________________________________________________________
up_sampling2d_1 (UpSampling2D) (None, 20, 20, 256) 0 add_2[0][0]
__________________________________________________________________________________________________
up_sampling2d (UpSampling2D) (None, 20, 20, 256) 0 batch_normalization_8[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 20, 20, 256) 65792 up_sampling2d_1[0][0]
__________________________________________________________________________________________________
add_3 (Add) (None, 20, 20, 256) 0 up_sampling2d[0][0]
conv2d_4[0][0]
__________________________________________________________________________________________________
activation_9 (Activation) (None, 20, 20, 256) 0 add_3[0][0]
__________________________________________________________________________________________________
conv2d_transpose_2 (Conv2DTrans (None, 20, 20, 128) 295040 activation_9[0][0]
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 20, 20, 128) 512 conv2d_transpose_2[0][0]
__________________________________________________________________________________________________
activation_10 (Activation) (None, 20, 20, 128) 0 batch_normalization_9[0][0]
__________________________________________________________________________________________________
conv2d_transpose_3 (Conv2DTrans (None, 20, 20, 128) 147584 activation_10[0][0]
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 20, 20, 128) 512 conv2d_transpose_3[0][0]
__________________________________________________________________________________________________
up_sampling2d_3 (UpSampling2D) (None, 40, 40, 256) 0 add_3[0][0]
__________________________________________________________________________________________________
up_sampling2d_2 (UpSampling2D) (None, 40, 40, 128) 0 batch_normalization_10[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 40, 40, 128) 32896 up_sampling2d_3[0][0]
__________________________________________________________________________________________________
add_4 (Add) (None, 40, 40, 128) 0 up_sampling2d_2[0][0]
conv2d_5[0][0]
__________________________________________________________________________________________________
activation_11 (Activation) (None, 40, 40, 128) 0 add_4[0][0]
__________________________________________________________________________________________________
conv2d_transpose_4 (Conv2DTrans (None, 40, 40, 64) 73792 activation_11[0][0]
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 40, 40, 64) 256 conv2d_transpose_4[0][0]
__________________________________________________________________________________________________
activation_12 (Activation) (None, 40, 40, 64) 0 batch_normalization_11[0][0]
__________________________________________________________________________________________________
conv2d_transpose_5 (Conv2DTrans (None, 40, 40, 64) 36928 activation_12[0][0]
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 40, 40, 64) 256 conv2d_transpose_5[0][0]
__________________________________________________________________________________________________
up_sampling2d_5 (UpSampling2D) (None, 80, 80, 128) 0 add_4[0][0]
__________________________________________________________________________________________________
up_sampling2d_4 (UpSampling2D) (None, 80, 80, 64) 0 batch_normalization_12[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 80, 80, 64) 8256 up_sampling2d_5[0][0]
__________________________________________________________________________________________________
add_5 (Add) (None, 80, 80, 64) 0 up_sampling2d_4[0][0]
conv2d_6[0][0]
__________________________________________________________________________________________________
activation_13 (Activation) (None, 80, 80, 64) 0 add_5[0][0]
__________________________________________________________________________________________________
conv2d_transpose_6 (Conv2DTrans (None, 80, 80, 32) 18464 activation_13[0][0]
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 80, 80, 32) 128 conv2d_transpose_6[0][0]
__________________________________________________________________________________________________
activation_14 (Activation) (None, 80, 80, 32) 0 batch_normalization_13[0][0]
__________________________________________________________________________________________________
conv2d_transpose_7 (Conv2DTrans (None, 80, 80, 32) 9248 activation_14[0][0]
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 80, 80, 32) 128 conv2d_transpose_7[0][0]
__________________________________________________________________________________________________
up_sampling2d_7 (UpSampling2D) (None, 160, 160, 64) 0 add_5[0][0]
__________________________________________________________________________________________________
up_sampling2d_6 (UpSampling2D) (None, 160, 160, 32) 0 batch_normalization_14[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 160, 160, 32) 2080 up_sampling2d_7[0][0]
__________________________________________________________________________________________________
add_6 (Add) (None, 160, 160, 32) 0 up_sampling2d_6[0][0]
conv2d_7[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 160, 160, 3) 867 add_6[0][0]
==================================================================================================
Total params: 2,058,979
Trainable params: 2,055,203
Non-trainable params: 3,776
__________________________________________________________________________________________________
结构图
netron.app 示例
