This repo is implementation for PointNet(https://arxiv.org/abs/1612.00593) in pytorch. The model is in pointnet.py.

conda env create -f environment.yml # create environment
source activate pointnet_pytorch # activate installed environment

1. Download data here
2. Extract to pointnet.pytorch project folder, we will obtain a folder of data named shapenetcore_partanno_segmentation_benchmark_v0

python train_segmentation.py # train 3D model segmentaion

bash build.sh # build C++ code for visualization
python show_seg.py --model seg/seg_model_1.pth --idx 5 # show segmentation results

Without heavy tuning, PointNet can achieve 80-90% performance in classification and segmentaion on this dataset.

Sample segmentation result:

• Project Page
• Tensorflow implementation