SWCF-Net

SWCF-Net is accepted by IROS 2024 🎉🎉🎉

Environment

The code has been tested on Ubuntu 20.04 with 1 Nvidia 3090 GPU (24GB memory).

1. Python 3.8.16

   conda create -n your_env_name python=3.8.16 -y

2. Install torch 1.10.0 + cu113

   pip install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio==0.10.0+cu113

3. nearest_neighbors && cpp_wrappers

   cd utils/nearest_neighbors
   python setup.py install --home="."
   cd ../../

   cd utils/cpp_wrappers
   sh compile_wrappers.sh
   cd ../../../

SemanticKITTI Segmentation

1. Data: SemanticKITTI dataset can be found here. Download the files related to semantic segmentation and extract everything into the same folder. Uncompress the folder and move it to /data/semantic_kitti/dataset. Run the following command to prepare the dataset.

   python data_prepare.py --src_path your_dataset_path --dst_path your_process_path

2. Train:

   python train_SemanticKITTI.py --log_dir your_result_path

Citation

If you find SWCF-Net useful to your research, please cite our work as an acknowledgment.

@article{lin2024swcf,
  title={SWCF-Net: Similarity-weighted Convolution and Local-global Fusion for Efficient Large-scale Point Cloud Semantic Segmentation},
  author={Lin, Zhenchao and He, Li and Yang, Hongqiang and Sun, Xiaoqun and Zhang, Cuojin and Chen, Weinan and Guan, Yisheng and Zhang, Hong},
  journal={arXiv preprint arXiv:2406.11441},
  year={2024}
}