affinelayer / pix2pix-tensorflow

Tensorflow port of Image-to-Image Translation with Conditional Adversarial Nets https://phillipi.github.io/pix2pix/
MIT License
5.07k stars 1.3k forks source link

How to use .tfrecords to load data from memory #166

Open ammar7971 opened 5 years ago

ammar7971 commented 5 years ago

I am working on my problem where I would like to add the classification loss in the generator. For that case, I need to change the way this code loads in the data from disk. I am thinking to use .tfrecords to load my images and corresponding labels in memory.

The problem I am facing is how to modify the current training procedure as right now all the fetches run at once using sess obj and the progress is calculated in a loop. But what I want is to first load a batch of images (using .tfrecords) and then feed the data to the network.

I am new in Tensorflow and GANS so finding it hard. Any suggestions or related fork will be appreciated. Thanks.