CodedVTR: Codebook-based Sparse Voxel Transformer with Geometric Guidance(CVPR22)

Tianchen Zhao, Niansong Zhang, Xuefei Ning, He Wang, Li Yi*, Yu Wang

The code implemention of codedvtr, for motr information, please check our project page: https://a-suozhang.xyz/codedvtr.github.io/

Installation

• please refer to the setup.md

Example

to train the model on semantic-kitti

./kitti-train.sh $EXP_NAME $GPU_ID

for multi-gpu training

./kitti-train-mp.sh test 0,1

Acknowledgment

This repository is developed based on chrischoy/SpatioTemporalSegmentation-ScanNet

Citation

if you find this project useful in your research, please consifer citing:

@inproceedings{zhao2022codedvtr,
  title={CodedVTR: Codebook-based Sparse Voxel Transformer with Geometric Guidance},
  author={Zhao, Tianchen and Zhang, Niansong and Ning, Xuefei and Wang, He and Yi, Li and Wang, Yu},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={1435--1444},
  year={2022}
}