Journal
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
Volume -, Issue -, Pages 10787-10797Publisher
IEEE
DOI: 10.1109/ICCV48922.2021.01063
Keywords
-
Funding
- Snap Inc. [EP/W002981/1]
- EPSRC/MURI [EP/N019474/1]
- EPSRC [EP/N019474/1] Funding Source: UKRI
Ask authors/readers for more resources
Full-motion cost volumes are crucial in optical flow methods, but their lack of encapsulating prior knowledge leads to artifacts in ambiguous regions. Our proposed separable cost volume module effectively utilizes global context clues and prior knowledge to accurately identify motions in these areas.
Full-motion cost volumes play a central role in current state-of-the-art optical flow methods. However, constructed using simple feature correlations, they lack the ability to encapsulate prior, or even non-local knowledge. This creates artifacts in poorly constrained ambiguous regions, such as occluded and textureless areas. We propose a separable cost volume module, a drop-in replacement to correlation cost volumes, that uses non-local aggregation layers to exploit global context cues and prior knowledge, in order to disambiguate motions in these regions. Our method leads both the now standard Sintel and KITTI optical flow benchmarks in terms of accuracy, and is also shown to generalize better from synthetic to real data.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available