DVG-Face icon indicating copy to clipboard operation
DVG-Face copied to clipboard

[TPAMI 2021] DVG-Face: Dual Variational Generation for Heterogeneous Face Recognition

Results 5 DVG-Face issues
Sort by recently updated
recently updated
newest added

论文中提到的判别器和对抗生成loss在代码中没有看到相应的实现? ``` # all losses loss = loss_ort + loss_pix + loss_ip ``` which is inconsistent with ![temp](https://user-images.githubusercontent.com/51785266/200517161-c17f0402-a40e-4296-8c2c-9cd53acf3526.png)

Thank you for your great work! I have a question about your paper. In your paper, you said that "Identity Sampler is the decoder of VAE". However, there isn't implementation...

I don't understand on what basis the labeling of the images is done in the train list(sampled data). Please help.

What's your test method to get the good results? In the code of provided, gallery set isn't used, as I know, the result on probe set only cannot represent Heterogeneous...

I don't understand how to fill out the --train_list parameter. Is it to save the example to a txt file and then fill in the path of the txt file?