fenghe12
fenghe12
用豆瓣源可以解决
Thanks! but i can't download from your onedrive link .Is there anything wrong with it?
ok thanks a lot！ By the way,i'm also studying emo,maybe I can learn from your repo
I cloned this repository yesterday,but I guess there is some bugs in training code(e.g. train_stage_1.py) Will anyone try this code and tell me how to run the pipeline?Thanks a lot！
有人帮忙看看上面的仓库吗，国外的一个作者复现了EMO，我大致看了一下代码没有很大的问题，但是现在训练脚本跑不起来，哪位大神帮忙看看！
其实问题挺大的，我只看了架子。细节基本是漏洞百出
> from hey can you share your downloaded dataset with me？thanks a lot~
为什么要采用Reference Image的信息用Spatial-Attention和Cross-Attention编码到Pose Sequence的Denoising UNet里面的方法而不是反过来
没有什么数学原理吧 就是效果更好 现在没办法很好地从数学上解释
if you’re trainging it,you may need more than 52GB vram
Have you finished this？ I want to visualize keypoints too