Awesome
Regularizing Self-supervised 3D Scene Flows with Surface Awareness and Cyclic Consistency
<p align="center"> <img src='docs/overview-merge.png'/> <p> <!-- # Results on StereoKITTI dataset --> <!--  -->Installation
- Install Fast Geodis with pip install FastGeodis --no-build-isolation
- Install PyTorch3d with CUDA support.
- Run commands in install.sh for installation of the packages above
DATA
- Setup directory for extracting the data, visuals and experimental results
export BASE_PATH='path_where_to_store_data'
- Download Data and unpack it to the folder $BASE_PATH/data/sceneflow:
tar -xvf data_sceneflow.tgz $BASE_PATH/data/sceneflow
Run Experiments
To run the method on all datasets with final metrics printed on cuda:0, just type:
for i in {0..6}; do python evaluate_flow.py $i; done
where the argument sets the specific datasets according to the following table:
| Dataset | Argument Number | Model |
|---|---|---|
| KITTI t | 0 | Neural Prior |
| StereoKITTI | 1 | Neural Prior |
| KITTI t | 2 | SCOOP |
| StereoKITTI | 3 | SCOOP |
| Argoverse | 4 | Neural Prior |
| Nuscenes | 5 | Neural Prior |
| Waymo | 6 | Neural Prior |
Experimental results on LiDAR Datasets
<!-- <img style="float: ;" src="docs/table_lidar.png"> --> <p align="center"> <img src="docs/table_lidar.png" /> </p>Experimental results on StereoKITTI Dataset
<p align="center"> <img src="docs/table_kitti.png" /> </p>Qualitative Example
