Closed amiltonwong closed 4 years ago
Hi,
The code used for the paper was messy and also written in TensorFlow. I've pushed a reimplemented version and verified that the default hyperparameters roughly reproduce the results in the paper. You'll have to obtain the processed dataset (ModelNet40_cloud.h5) separately since we cannot publicly share it.
Thanks, Yoonho Lee
Hi, @juho-lee ,
Thanks for releasing this package. Where could I find the code for modelnet 40 shape classification as mentioned in the paper?
THX!