Learning Dense Correspondences for Non-Rigid Point Clouds With Two-Stage Regression

paper

Network

Installation

Please refer requirements.txt to prepare the environment.

Then you need compile the code in */tf_ops/. First enter the corresponding folder, i.e. global_regression/tf_ops/3d_interpolation

For Stage1 and Stage2, run

make clean && make

For Finetune Stage1 and Finetune Stage2, run

./tf_interpolate_compile.sh

Question

If you have any questions about compiling /tf_ops/ when running the code, please first look for solutions from pointnet2.

Run the code

Please refer the corresponding README.md

Stage1 and Stage2

Finetune Stage1 and Finetune Stage2

Citation

If you find this code useful for your research, please use the following BibTeX entry.

@ARTICLE{Wang_learnging,  
  title={Learning Dense Correspondences for Non-Rigid Point Clouds With Two-Stage Regression},   
  author={Wang, Kangkan and Zhang, Guofeng and Zheng, Huayu and Yang, Jian},    
  journal={IEEE Transactions on Image Processing},   
  year={2021},  
  volume={30},  
  pages={8468-8482},  
  doi={10.1109/TIP.2021.3116786}
}