README.md

AGMMConv

This repository is the implementation of "Adaptive GMM Convolution for Point Cloud Learning"

1. Setup

1) Building

cd utils
sh complile_op.sh

2) Dependency

This repository is partially dependent on 'pytorch', 'torch_geometric' and 'torch_points3d'.

2. Runing

Please see 'trainval.y' for classification and segmentation tasks, and 'trainval_normal.py' for normal estimation task.
Please see 'configure.py' for the settings.
And see the 'dataset' directories for the data processing.

3. Acknowledgement

Part of the codes refers to the KPConv and RandLA-Net

4. Others

This repository will continue updating.