waxnkw/gvcnn-pytorch

A PyTorch version group view convolution network. It is used in 3D classification and 3D retrieval.

GVCNN-Pytorch Version

Environment: python 3.6 PyTorch 1.0.0

The code is implemented on the base of jongchyisu's mvcnn_pytorch, and is inspired by Sean Kim's tensorflow version.

The inception v4 can be found here.

Paper

The paper can be found here.

Modification

There are two main difference from the origin paper:

1. Considering the huge amount of fc layer, we use 1*1 conv instead of fc in group schema module.
2. If all views' scores are very small, it may cause some problem in params' update. So we add a softmax to normalize the scores generate by group schema.

Run

modify root in data_process.py to your ModelNet40 dataset.

python data_process.py

python gvcnn_train.py

Reference

• Feng Y, Zhang Z, Zhao X, et al. GVCNN: Group-view convolutional neural networks for 3D shape recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2018: 264-272.

• https://github.com/jongchyisu/mvcnn_pytorch

• https://github.com/ace19-dev/gvcnn

• https://github.com/Cadene/pretrained-models.pytorch