FaceX-Zoo icon indicating copy to clipboard operation
FaceX-Zoo copied to clipboard

A PyTorch Toolbox for Face Recognition

Results 120 FaceX-Zoo issues
Sort by recently updated
recently updated
newest added

你好,使用RAF-DB数据集训练网络时参数设置与AffectNet数据集相同吗?代码中设置了两个学习率lr1和lr2,但论文中只提到了一个学习率,请问论文中的学习率指的是哪个。代码中的学习率与论文中提到的设置并不相同,且关于ramp_up的公式设置也不相同,我在RAF-DB数据集上尝试了代码中与论文中的设置方式,均无法达到论文中所提到的精度,能否告知在RAF-DB和FERPlus数据集上训练时参数设置,感谢!

你好,我对DSDG中测试代码([FaceX-Zoo](https://github.com/JDAI-CV/FaceX-Zoo)/[addition_module](https://github.com/JDAI-CV/FaceX-Zoo/tree/main/addition_module)/[DSDG](https://github.com/JDAI-CV/FaceX-Zoo/tree/main/addition_module/DSDG)/[DUM](https://github.com/JDAI-CV/FaceX-Zoo/tree/main/addition_module/DSDG/DUM)/test.py)有点疑问,请问测试中计算score的时候,有用到norm的操作(score_norm = torch.sum(mu) / torch.sum(test_maps[:, frame_t, :, :])),针对这个我有两个疑惑: 1、假如是一个fake的样本,torch.sum(test_maps[:, frame_t, :, :]应该等于0?那是否需要加一个偏置项来避免除0的情况发生? 2、假如网络训练的很好的话,mu和test_maps[:, frame_t, :, :]应近似相等?那不论是对real还是fake的样本,score_norm应该都近似为1吧?怎么在计算指标的时候对他们进行区分呢

face_masker.py image = imread(image_path)

face_masker.py uv_face = imread(self.uv_face_path, as_gray=True)/255. how to remove the warning message?

Hi, Thanks for sharing, Could you share the snip code to predict face is Fake or Not with CDCN_U_P1.pkl trained model. Wait for you response. Thank you.

请问哪里可以获取到MS-Celeb-1M-v1c数据集呢,是否是可开放的数据集?

While running add_mask_one.py, I got an error "**ValueError: Buffer dtype mismatch, expected 'long' but got 'long long'**", it shows the render_cy in line 165 of face_masker.py and the render_cy in...