Skip to content

Latest commit

 

History

History
31 lines (27 loc) · 719 Bytes

README.md

File metadata and controls

31 lines (27 loc) · 719 Bytes

SceneFlowNet

Implementation of 3D SceneFlowNet: Self-Supervised 3D Scene Flow Estimation

Install the dependencies as:

conda dependencies/environment.yml

Training SceneFlowNet

./run_flow_training.sh

Testing SceneFlowNet

./run_flow_testing.sh

Unzip model.best.zip under checkpoints/ to use pretrained model

Citation

If you find this useful, please cite as:

@inproceedings{lu20213d,
  title={3d sceneflownet: Self-supervised 3d scene flow estimation based on graph cnn},
  author={Lu, Yawen and Zhu, Yuhao and Lu, Guoyu},
  booktitle={2021 IEEE International Conference on Image Processing (ICIP)},
  pages={3647--3651},
  year={2021},
  organization={IEEE}
}