search

bcmi / DCI-VTON-Virtual-Try-On

[ACM Multimedia 2023] Taming the Power of Diffusion Models for High-Quality Virtual Try-On with Appearance Flow.
https://arxiv.org/abs/2308.06101
MIT License
415 stars 62 forks source link

poor evaluation metrics score #37

Open davvvy opened 10 months ago

davvvy commented 10 months ago

Hi,

Thanks for your work first!

I run inference by your checkpoint and evaluate all the metrics you mentioned in the paper but I got very poor results. Could you please provide the code for evaluating the performance?

Jerome-Young commented 10 months ago

Hi, I encountered the same issue.

Limbor commented 10 months ago

Hi, we tested the SSIM and LPIPS according to hr-vton’s evaluate.py, and the FID and KID metrics were obtained through the torch-fidelity library.

davvvy commented 10 months ago

Hi, we tested the SSIM and LPIPS according to hr-vton’s evaluate.py, and the FID and KID metrics were obtained through the torch-fidelity library.

Thanks for your reply! Also, I'm wondering how you evaluate the metrics at different image sizes. Do you train at different image sizes?

Limbor commented 10 months ago

Yes, we trained the model at three resolutions.