如何使用 tf.keras 从 savedModel 访问图层

Question

张量流 2.0

python 3.7

我使用 tf.keras

训练并保存了这样的模型

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import datasets, layers, models 
from tensorflow.keras.datasets import mnist

(train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()

# Normalize pixel values to be between 0 and 1
train_images, test_images = train_images / 255.0, test_images / 255.0

class_names = ['airplane', 'automobile', 'bird', 'cat', 'deer',
               'dog', 'frog', 'horse', 'ship', 'truck']

model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.Flatten())
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(10))

model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=['accuracy'])

history = model.fit(train_images, train_labels, epochs=10, 
                    validation_data=(test_images, test_labels))

test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)

model.save("saved_model_pretrained/")

我想加载模型以便在像这样的另一个项目中从中提取图层

image_model = tf.keras.models.load_model('pathToFolder')

我正在尝试获得这样的图层：

layer_indices = []
for index, layer in enumerate(image_model.layers):
    layers_indices.append(index)

但是我收到了这个错误

future: <Task finished coro=<server_task.<locals>.server_work() done, defined at ...\...\xx.py:249> exception=AttributeError("'_UserObject' object has no attribute 'layers'")>
Traceback (most recent call last):
  File "...\...\xx.py", line 280, in server_work
    image_model, layers_indices = init(model_choice, layers_to_see)
  File "...\...\xx.py", line 148, in init
    for index, layer in enumerate(image_model.layers):
AttributeError: '_UserObject' object has no attribute 'layers'

非常感谢任何帮助

Answer 1

首先，您应该妥善保存您的模型！使用以下语法将 save/load 模型作为 (.h5) 文件 - 保存模型的常用格式：

model.save("model.h5")
image_model = tf.keras.models.load_model('model.h5')

那么你的 image_model 里面就会有 layers。在您的情况下，image_model.layers 产生：

[<tensorflow.python.keras.layers.convolutional.Conv2D at 0x1d81b325898>,
 <tensorflow.python.keras.layers.pooling.MaxPooling2D at 0x1d81b325f98>,
 <tensorflow.python.keras.layers.convolutional.Conv2D at 0x1d82b06b2e8>,
 <tensorflow.python.keras.layers.pooling.MaxPooling2D at 0x1d82c1dbe10>,
 <tensorflow.python.keras.layers.convolutional.Conv2D at 0x1d82c17d2e8>,
 <tensorflow.python.keras.layers.core.Flatten at 0x1d82c17df98>,
 <tensorflow.python.keras.layers.core.Dense at 0x1d82c2c8320>,
 <tensorflow.python.keras.layers.core.Dense at 0x1d82b016048>]

和

layers_indices = []
for index, layer in enumerate(image_model.layers):
    layers_indices.append(index)
layers_indices

产生：

[0, 1, 2, 3, 4, 5, 6, 7]

如何使用 tf.keras 从 savedModel 访问图层

How to access layers from a savedModel with tf.keras

python

keras

tensorflow

tf.keras

tensorflow2.0