You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
2、关于损失的计算,论文中的3D reconstruction losses:$L _ { r e c } = | | \widetilde { O } _ { s } ( p ) - O _ { g t } ( p ) | | ^ { 2 }$
其中$\widetilde { O } _ { s } ( p )$是Fine Occ(Coarse occ+Δocc)。但是在代码中是使用
你好,看来你的代码,有几个疑问的地方
1、关于 MLP_DIF 代码中的self.filters_fine,这个MLP网络的输出值,fine_y 对应的是论文中的Δ Occ还是Fine Occ?
2、关于损失的计算,论文中的3D reconstruction losses:$L _ { r e c } = | | \widetilde { O } _ { s } ( p ) - O _ { g t } ( p ) | | ^ { 2 }$ $\widetilde { O } _ { s } ( p )$ 是Fine Occ(Coarse occ+Δocc)。但是在代码中是使用
其中
https://github.com/psyai-net/D-IF_release/blob/f5b82ff5e18ca42115741c3f808520bf664329d0/lib/net/HGPIFuNet.py#L403C1-L405C60
看起来是将第二个MLP(Occupancy Rectifier)的输出pred_if和第一个MLP(Distribution Predictor)的输出之一$\mu$ 都做了一次L2损失,并相加,这看起来和论文中似乎不一致。
3、最后,train_step中,对损失乘以0了,这是为什么?
D-IF_release/apps/ICON.py
Line 217 in f5b82ff
@yxt7979
The text was updated successfully, but these errors were encountered: