Which dataset should we use to calculate FID score?

[TianJiaJiaFirst]

I read on some resources that the FID score should be calculated using the dataset of generated outputs and the dataset that was used to train the generator. Which datasets should we use to get the FID score? It seems that comparing with the training dataset (which is much larger) I usually get a better FID score for CP-VTON. Following the file you gave, test_pairs.txt it seems that there is no grand truth to compare it.

[xxxxl888]

I read on some resources that the FID score should be calculated using the dataset of generated outputs and the dataset that was used to train the generator. Which datasets should we use to get the FID score? It seems that comparing with the training dataset (which is much larger) I usually get a better FID score for CP-VTON. Following the file you gave, test_pairs.txt it seems that there is no grand truth to compare it.

Can you tell me which two datasets should be used to calculate the FID score? I have encountered this problem too

[TianJiaJiaFirst]

I read on some resources that the FID score should be calculated using the dataset of generated outputs and the dataset that was used to train the generator. Which datasets should we use to get the FID score? It seems that comparing with the training dataset (which is much larger) I usually get a better FID score for CP-VTON. Following the file you gave, test_pairs.txt it seems that there is no grand truth to compare it.

Can you tell me which two datasets should be used to calculate the FID score? I have encountered this problem too

Sorry, I don't know either