search

nihaomiao / CVPR23_LFDM

The pytorch implementation of our CVPR 2023 paper "Conditional Image-to-Video Generation with Latent Flow Diffusion Models"
BSD 2-Clause "Simplified" License
432 stars 43 forks source link

FVD calculation #36

Closed take323 closed 7 months ago

take323 commented 8 months ago

Thanks for your work!

I used the pretrained models published in this repository and calculated the FVD using this repository. However, I obtained FVDs much larger than those in the paper. Could you please release the source code for calculating FVD?

HARVEYDOU commented 7 months ago

same question

nihaomiao commented 7 months ago

Hi, @take323 @HARVEYDOU , thanks a lot for your interest in our work! I think I use the same code as the repo you mentioned. But FVD depends on many factors, such as the number of videos (we use 10,000), the resolution, number of frames, and sample rate (we sample 40 frames with the resolution of 128*128 from the ground truth video), etc. Using different settings may result in different FVD scores.