Giters

bioinf-jku / TTUR

Two time-scale update rule for training GANs

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Question regarding FID Score

miqbal23 opened this issue · comments

commented

Hello, I'm interested in using your code to evaluate several GAN models that I used as of now. Although, I have few question regarding your implementation of FID score

  1. Recently I've been digging on implementation of FID, and found that Google released an evaluation paper using your method (code). However, they used Inception V3 while you still used Inception V1 here. Which one should I use?
  2. From your code, I found that building the pretrained model using our data also need batch_size parameter. How does this affect the calculation? And which size is better for a given set of data?
commented

Hi, thanks for using our code.

  1. Can you give us a direct link where in the Google code they used Inception V3?

  2. The parameter batch_size, as described in the code, splits the image data array in junks to propagate through the inception net. The calculation is not affected as the code collects the batched results (inception net activations) into one final array, one row for one image, to calculate the statistics.

commented

Hi!
The compare_gan code uses Inception V1, as they use the code from this TF module (which clearly uses V1): https://github.com/tensorflow/tensorflow/blob/r1.10/tensorflow/contrib/gan/python/eval/python/classifier_metrics_impl.py#L70