GDS

This repo is for paper GDS:Global Description Guided Down-Sampling for 3D Point Cloud Classification

Requirements

• Python 3.7
• PyTorch 1.2
• CUDA 10.0
• Package: glob, h5py, sklearn

Point Cloud Classification

ModelNet

Run the training script:

python main_cls.py --base_dir $Your Base Dir$ --exp_name=cls_1024 --num_points=1024 --k=20 --cd_weights 0.1

Run the evaluation script after training finished:

python main_cls.py --base_dir $Your Base Dir$ --exp_name=cls_1024_eval --num_points=1024 --k=20 --eval=True --model_path=$Your Model Path$

ScanObjectNN

Run the training script:

python main_scan.py --base_dir $Your Base Dir$ --exp_name=scan_bg_1024 --num_points=1024 --k=20 --cd_weights 1.0

Run the evaluation script after training finished:

python main_scan.py --base_dir $Your Base Dir$ --exp_name=scan_bg_1024_eval --num_points=1024 --k=20 --eval=True --model_path=$Your Model Path$

Citing GDS

If you find this repo useful, consider citing GDS use the following format,

ACM Reference format: Jiahua Wang, Yao Zhao, Ting Liu and Shikui Wei. 2020. GDS:Global Description Guided Down-Sampling for 3D Point Cloud Classification. In Proceedings of 2020 4th International Conference on Vision, Image and Signal Proceeding (ICVISP 2020), December 9-11, 2020, Bangkok, Thailand. ACM, New York, NY, USA, 7 pages.

Acknowledgement

This repo is based upon dgcnn.pytorch.