Laka-3DV/ML-SemReg

Boosting PCR with Multi-level Semantic Consistency

ML-SemReg

ML-SemReg: Boosting Point Cloud Registration with Multi-level Semantic Consistency

Shaocheng Yan¹, Pengcheng Shi², Jiayuan Li^{1 📧}

¹School of Remote Sensing and Information Engineering, Wuhan University, ²School of Computer Science, Wuhan University

(*) equal contribution, (^📧) corresponding author.

ML-SemReg is a new plug-and-play method for boosting point cloud registration, utilizing multi-level semantic consistency. Its core idea is to address inter- and intra-class mismatchings (outliers) ultilizing multi-level semantic consistency.

Demo

Installation

conda create -n mlsemreg python=3.9 
conda activate mlsemreg
pip install -r requirements.txt

# please check your CUDA version
pip install torch==1.13.1+cu117 torchvision==0.14.1+cu117 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu117

Run demo.py

# a demo using KITTI medium dataset
python -m demo
python -m demo -is_vis

Citation

If you use this codebase, or otherwise find our work valuable, please cite ML-SemReg:

TODO

Acknowledgements

• Pointcept
• SC2-PCR