Home

Awesome

FAL_net

Repository of our NeurIPS2020 Accepted paper "Forget About the LiDAR: Self-Supervised Depth Estimators with MED Probability Volumes" Paper and supplemental materials: https://proceedings.neurips.cc/paper/2020/hash/951124d4a093eeae83d9726a20295498-Abstract.html

Pre-trained models

Pre-trained model (KITTI-only, 1st stage)

Pre-trained model (KITTI-only, 2nd stage)

Computed inverse depth maps

Improved Eigen test split Trained on KITTI-only, 2nd_stage, and post-processing

Cite our paper:

@inproceedings{NEURIPS2020_951124d4,
 author = {Gonzalez Bello, Juan Luis and Kim, Munchurl},
 booktitle = {Advances in Neural Information Processing Systems},
 editor = {H. Larochelle and M. Ranzato and R. Hadsell and M. F. Balcan and H. Lin},
 pages = {12626--12637},
 publisher = {Curran Associates, Inc.},
 title = {Forget About the LiDAR: Self-Supervised Depth Estimators with MED Probability Volumes},
 url = {https://proceedings.neurips.cc/paper/2020/file/951124d4a093eeae83d9726a20295498-Paper.pdf},
 volume = {33},
 year = {2020}
}

New: Added files to train our FAL-net on KITTI

Coming soon: Files to train our FAL-net on KITTI + CityScapes

Not for commercial use If wanted for commercial use contact juanluisgb@kaist.ac.kr