arnoldfychen / SST

Codes for “Fully Sparse 3D Object Detection” & “Embracing Single Stride 3D Object Detector with Sparse Transformer”

This repo contains official implementations of our series of work in LiDAR-based 3D object detection:

• Embracing Single Stride 3D Object Detector with Sparse Transformer (CVPR 2022).
• Fully Sparse 3D Object Detection (NeurIPS 2022).
• Super Sparse 3D Object Detection

Users could follow the instructions to use this repo.

NEWS

• [23-03-29] 🔥 We develop a high-performance offline detector CTRL, preview the single-model performance in leaderboard.
• [23-03-21] 🔥 The Argoverse 2 model of FSD is released. See instructions.
• [22-09-19] The code of FSD is released here.
• [22-09-15] FSD is accepted at NeurIPS 2022.
• [22-06-06] Support SST with CenterHead, cosine similarity in attention, faster SSTInputLayer.
• [22-03-02] SST is accepted at CVPR 2022.
• [21-12-10] The code of SST is released.

Citation

Please consider citing our work as follows if it is helpful.

Since FSD is accidentally excluded in Google Scholar search results, if possible, please kindly consider citing the journal version of FSD as well.

@inproceedings{fan2022embracing,
  title={{Embracing Single Stride 3D Object Detector with Sparse Transformer}},
  author={Fan, Lue and Pang, Ziqi and Zhang, Tianyuan and Wang, Yu-Xiong and Zhao, Hang and Wang, Feng and Wang, Naiyan and Zhang, Zhaoxiang},
  booktitle={CVPR},
  year={2022}
}

@inproceedings{fan2022fully,
  title={{Fully Sparse 3D Object Detection}},
  author={Fan, Lue and Wang, Feng and Wang, Naiyan and Zhang, Zhaoxiang},
  booktitle={NeurIPS},
  year={2022}
}

@article{fan2023super,
  title={Super Sparse 3D Object Detection},
  author={Fan, Lue and Yang, Yuxue and Wang, Feng and Wang, Naiyan and Zhang, Zhaoxiang},
  journal={arXiv preprint arXiv:2301.02562},
  year={2023}
}

Acknowledgments

This project is based on the following codebases.

• MMDetection3D
• LiDAR-RCNN

Thank the authors of CenterPoint for providing their detailed results.