search

charlesq34 / pointnet

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
Other
4.66k stars 1.44k forks source link

Shared MLP code #160

Open powenlo opened 5 years ago

powenlo commented 5 years ago

I am still confused about why the shared MLP can be achieved by 1D convolution in the code. It should be FC layers, right? Can anyone explain this to me? (better to give me an example) Thanks a lot.

dashujuzha commented 5 years ago

l have the same,question.

heexu976 commented 2 years ago

I think convolution can do same things as MLP ，and convolution is easier to train.