terrychenism/OctaveConv

How does OctConv really implementstrided convolutions

rongmelt opened this issue · 1 comments

First of all, feel the author's open source spirit, great idea!

I know that low and high frequencies are divided according to a given proportion, but how are they divided along the channel dimension? Are they divided randomly according to proportion? Or are there other pre-processing in it?

Thank you very much.

从代码上看,感觉高频的是通过一层卷积后得到,低频的话,是先pooling 再卷积得到的