using a trained cnn for a test data

Question

I'm training a CNN in Tensorflow and I save it with the code:

saver = tf.train.Saver()

end of each Epoch I save it using this, saver.save(sess,Save_Dir).

After my CNN has been trained, I want to use it on a test data. I wonder, How I can restore my Trained CNN and How I can see How my trained CNN works on test data that I have. I've searched a lot in this regard, but I didn't find something that I can Learn about the process. Moreover, I do not have any sense about the 4 types of files that the saver generate. (.meta,....). Thank You.

Possible duplicate of [Tensorflow: how to save/restore a model?](https://stackoverflow.com/questions/33759623/tensorflow-how-to-save-restore-a-model) — E_net4, Oct 22 '18 at 09:15

score 0 · Accepted Answer · answered Oct 22 '18 at 14:43

restore my Trained CNN

To load your model, you do saver.restore(session, load_path) which should load the graph back into the session.

How I can see How my trained CNN works on test data that I have

You can use tf.metrics or define your own custom metric

Moreover, I do not have any sense about the 4 types of files that the saver generate. (.meta,....). Thank You.

I found Tensorflow: The confusing Parts 2 to be extremely useful.

using a trained cnn for a test data

1 Answers1