Semantic point cloud segmentation with graph convolutional network

Class project for CMU 10707 Deep Learning.

Usage

First, download data from https://shapenet.cs.stanford.edu/iccv17.

Then, install tensorpack and lmdb. Use write_lmdb.py to prepare data. For example, python3 write_lmdb.py train -c Airplane prepares training data for the Airplane category.

Use train.py to train a model and test.py to test it. Run python3 train.py -h to see available options. Segmentation results can be visualized using visualize_results.py.