chunyuwang / lcn-pose Goto Github PK

I can't test the model trained on Human3.6M on MPI-INF-3DHP, the result is very strange. I wonder if there are any special fine-turning skills or something? Could you please release the model on MPI-INF-3DHP and the corresponding test code? Thank you!

1.您实现的GCN是不需要用拉普拉斯矩阵吗？我不知道我的理解是否正确，您提出的式子y = X (S ⊙ W)，实际上在FCN里相当于y = X (1 ⊙ W)，在GCN里相当于y = X (A ⊙ W)，A是根据拉普拉斯矩阵算出来的东西。我不能理解的是，您的式子里S和W都是未知变量吗？这样的结构如何保证它能收敛呢？

2.我想了解下您的auc和pck是如何计算的，您可以开源一下在mpi-inf-3dhp数据集上的代码吗？

What is the purpose of confidence in the code, Can adding this dimension additionally improve accuracy? Will this bring additional computational burden?

Hello, thank you very much for your work. I want to know how your paper obtains the error value of each action? Thank you for your reply and wish you all the best.