Question

我目前正在尝试学习如何使用TF-Slim，我正在学习本教程：https://github.com/mnuke/tf-slim-mnist。

假设我已经在检查点保存了经过训练的模型，我现在如何使用该模型并应用它？就像在教程中一样，我如何使用训练有素的MNIST模型并输入一组新的MNIST图像，并打印预测？

Answer 1

您可以尝试以下工作流程：

#obtain the checkpoint file
checkpoint_file= tf.train.latest_checkpoint("./log")

#Construct a model as such:
with slim.arg_scope(mobilenet_arg_scope(weight_decay=weight_decay)):
            logits, end_points = mobilenet(images, num_classes = dataset.num_classes, is_training = True, width_multiplier=width_multiplier)

#Obtain the trainable variables and a saver
variables_to_restore = slim.get_variables_to_restore()
saver = tf.train.Saver(variables_to_restore)

#Proceed to create your training optimizer and predictions monitoring, summaries etc.
...

#Finally when you're about to train your model in a session, restore the checkpoint model to your graph first:

with tf.Session() as sess:
    saver.restore(sess, checkpoint_file)
    #...Continue your training

基本上，您必须获取要恢复的正确变量，并且这些变量必须具有与检查点模型中找到的名称相匹配的名称。然后，将要恢复的变量列表传递给Saver，然后在TF会话中，让saver从会话中的检查点模型中恢复所有变量。

Answer 2

查看官方TF-Slim documentation和walkthrough

Tensorflow Slim恢复模型和预测

2 个答案: