landmark loss decreased slowly

Open anewusername77 opened this issue 3 years ago • 0 comments

Hey there, thanks for your impressive work! I tried Nextface first and now implemented DeepNextFace by myself, using default resnet152 in pytorch(pretrained on ImageNet), but during step 1, landmark loss decreased in a quite slow way, ended up with about 40000. I think it may related to slow fitting of focal and camera position params. Dosen't find more details about the training strategy in your paper including appendix, can you please help?

Aug 29 '22 08:08 anewusername77