Naiyuan Liu

BTW, when I was running preprocess_video.py, I meet a bug : "RuntimeError: Cache file does not exist: E:\\Dataset\\IJB\\IJB-C\\images\\video\10016\10016_dsfd_seq.pkl"

Thank you for sharing the code!!!

I got some problem during training: the GPU memory usage keep rising during training. Here is the log information: [01/07 16:42:57 d2.utils.events]: eta: 6:37:20 iter: 339 total_loss: 3.58 loss_fcos_cls: 1.22...

Sorry for the late respond ,did you follow the installation section in readme.md to build your environment? And what kind of GPU did you use.

Oh, I suggest you to use CUDA >=11.1,because the defualt cuda in the installation instructions is 10.1 which is not suitable for 30 serial GPU

I meet the same problem, it is really hard to download this dataset in china. Did you find a way to solve it ? @lihuinian

Btw, as you mentioned in the paper,you use the stacked hourglass networks to extract landmarks. But I found the oringal implementation in the "Stacked hourglass networks for human pose estimation"...

May I ask whether the stacked hourglass of the structure extraction module in dfvae will be trained together, that is, whether the parameters of the stacked hourglass will undergo gradient...

Thank you for your quick answer! BTW, in appendix, you mentioned using kl loss as one of the training losses of dfvae. Does the kl loss here mean that the...

Could you further explain how the df-vae combined the structure imformation and appearance and how did the mean and variance in this part of klloss come from? I reproduced your...