Find a file
2021-10-05 16:49:00 -04:00
classification_ModelNet40 update 2021-10-04 04:00:01 -04:00
classification_ScanObjectNN update 2021-10-04 03:54:06 -04:00
part_segmentation update 2021-10-04 12:50:22 -04:00
pointnet2_ops_lib update 2021-10-04 03:25:18 -04:00
.gitignore update 2021-10-04 03:45:09 -04:00
LICENSE Initial commit 2021-10-04 03:01:38 -04:00
overview.pdf Add files via upload 2021-10-05 16:45:11 -04:00
overview.png Add files via upload 2021-10-05 16:49:00 -04:00
README.md Update README.md 2021-10-05 16:46:48 -04:00
requirements.txt update 2021-10-04 03:45:09 -04:00

pointMLP-pytorch

Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework

Overview of one stage in PointMLP. Given an input point cloud, PointMLP progressively extract local features using residual point MLP blocks. In each stage, we first transform local point using a geometric affine module, then local points are are extracted before and after aggregation respectively. By repeating multiple stages, PointMLP progressively enlarge the receptive field and model entire point cloud geometric information.

Pre-trained models

Please download the pre-trained models and log files here: [anonymous google drive]

Install

Please ensure that python3.7+ is installed. We suggest user use conda to create a new environment.

Install dependencies

pip install -r requirements.txt

Install CUDA kernels

pip install pointnet2_ops_lib/.

Classification ModelNet40

The dataset will be automatically downloaded, run following command to train

# train pointMLP
python main.py --model pointMLP
# train pointMLP-elite
python main.py --model pointMLPElite
# please add other paramemters as you wish.

By default, it will create a fold named "checkpoints/{modelName}-{msg}-{randomseed}", which includes args.txt, best_checkpoint.pth, last_checkpoint.pth, log.txt, out.txt.

To conduct voting experiments, run

# please modify the msg accrodingly
python voting.py --model pointMLP --msg demo

Classification ScanObjectNN

  • Make data folder and download the dataset
cd pointMLP-pytorch/classification_ScanObjectNN
mkdir data
cd data
wget http://103.24.77.34/scanobjectnn/h5_files.zip
unzip h5_files.zip
  • Train pointMLP/pointMLPElite
# train pointMLP
python main.py --model pointMLP
# train pointMLP-elite
python main.py --model pointMLPElite
# please add other paramemters as you wish.

By default, it will create a fold named "checkpoints/{modelName}-{msg}-{randomseed}", which includes args.txt, best_checkpoint.pth, last_checkpoint.pth, log.txt, out.txt.

  • To conduct voting experiments
# please modify the msg accrodingly
python voting.py --model pointMLP --msg demo

Part segmentation

  • Make data folder and download the dataset
cd pointMLP-pytorch/part_segmentation
mkdir data
cd data
wget https://shapenet.cs.stanford.edu/media/shapenetcore_partanno_segmentation_benchmark_v0_normal.zip --no-check-certificate
unzip shapenetcore_partanno_segmentation_benchmark_v0_normal.zip
  • Train pointMLP
# train pointMLP
python main.py --model pointMLP
# please add other paramemters as you wish.