PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume

License

Copyright (C) 2018 NVIDIA Corporation. All rights reserved. Licensed under the CC BY-NC-SA 4.0 license (https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode).

Usage

For Caffe users, please refer to Caffe/README.md.

For PyTorch users, please refer to PyTorch/README.md

The PyTorch implementation almost matches the Caffe implementation (average EPE on the final pass of the Sintel training set: 2.31 by Pytorch and 2.29 by Caffe).

Network Architecture

PWC-Net fuses several classic optical flow estimation techniques, including image pyramid, warping, and cost volume, in an end-to-end trainable deep neural networks for achieving state-of-the-art results.

Paper & Citation

Deqing Sun, Xiaodong Yang, Ming-Yu Liu, and Jan Kautz. "PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume." CVPR 2018 or arXiv:1709.02371

Updated and extended version: "Models Matter, So Does Training: An Empirical Study of CNNs for Optical Flow Estimation." arXiv:1809.05571

Project page link

Talk at robust vision challenge workshop

Talk at CVPR 2018 conference

If you use PWC-Net, please cite the following paper:

@InProceedings{Sun2018PWC-Net,
  author    = {Deqing Sun and Xiaodong Yang and Ming-Yu Liu and Jan Kautz},
  title     = {{PWC-Net}: {CNNs} for Optical Flow Using Pyramid, Warping, and Cost Volume},
  booktitle = CVPR,
  year      = {2018},
}

or the arXiv paper

@article{sun2017pwc,
  author={Sun, Deqing and Yang, Xiaodong and Liu, Ming-Yu and Kautz, Jan},
  title={{PWC-Net}: {CNNs} for Optical Flow Using Pyramid, Warping, and Cost Volume},
  journal={arXiv preprint arXiv:1709.02371},
  year={2017}
}

or the updated and extended version

@article{Sun2018:Model:Training:Flow,
  author={Sun, Deqing and Yang, Xiaodong and Liu, Ming-Yu and Kautz, Jan},
  title={Models Matter, So Does Training: An Empirical Study of CNNs for Optical Flow Estimation},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)},
  note = {to appear}
}

For multi-frame flow, please also cite

@inproceedings{ren2018fusion,
  title={A Fusion Approach for Multi-Frame Optical Flow Estimation},
  author={Ren, Zhile and Gallo, Orazio and Sun, Deqing and Yang, Ming-Hsuan and Sudderth, Erik B and Kautz, Jan},
  booktitle={Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV)},
  year={2019}
}

Related Work from NVIDIA

flownet2-pytorch

Learning Rigidity in Dynamic Scenes with a Moving Camera for 3D Motion Field Estimation (ECCV 2018)

Contact

Deqing Sun (deqing.sun@gmail.com)