1. [Publications](/index.php/publications)
2. A Fusion Approach for Multi-Frame Optical Flow Estimation
 
 # A Fusion Approach for Multi-Frame Optical Flow Estimation

  ![](/sites/default/files/styles/wide/public/publications/teaser_eraseme_0.png?itok=Nzuhp1jO)

 To date, top-performing optical flow estimation methods only take pairs of consecutive frames into account. While elegant and appealing, the idea of using more than two frames has not yet produced state-of-the-art results. We present a simple, yet effective fusion approach for multi-frame optical flow that benefits from longer-term temporal cues. Our method first warps the optical flow from previous frames to the current, thereby yielding multiple plausible estimates. It then fuses the complementary information carried by these estimates into a new optical flow field. At the time of writing, our method ranks first among published results in the MPI Sintel and KITTI 2015 benchmarks.



 ## Authors



Zhile Ren (Georgia Tech)

Orazio Gallo (NVIDIA)

Deqing Sun (NVIDIA)

Ming-Hsuan Yang (UC Merced)

Erik B. Sudderth (UC Irvine)

[Jan Kautz](/index.php/person/jan-kautz)

 

 

 ## Publication Date



Tuesday, January 8, 2019

 

 ## Published in



[IEEE Winter conference of Applications of Computer Vision (WACV)](http://wacv19.wacv.net/)

 

 ## Research Area



[Artificial Intelligence and Machine Learning ](/index.php/research-area/machine-learning-artificial-intelligence)

[Computer Vision](/index.php/research-area/computer-vision)

 

 

 ## External Links



[Paper](https://arxiv.org/pdf/1810.10066)

[Code](https://github.com/NVlabs/PWC-Net/tree/master/Multi_Frame_Flow)

 

 

 ## Uploaded Files



[Bibtex citation](https://research.nvidia.com/sites/default/files/pubs/2019-01_A-Fusion-Approach/citation.txt "Open document in new window")327 bytes

[Network weights (see code repo)](https://research.nvidia.com/sites/default/files/pubs/2019-01_A-Fusion-Approach/pwc_net.pth_.tar_.zip "Open archive in new window")32.14 MB

[Training/testing split for Monkaa and Virtual Kitti](https://research.nvidia.com/sites/default/files/pubs/2019-01_A-Fusion-Approach/vkitti_monkka_split.zip "Open archive in new window")940.09 KB

 

 

 ## Copyright



This material is posted here with permission of the IEEE. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to <pubs-permissions@ieee.org>.