Fast Estimation of Large Displacement Optical Flow using Dominant Motion Patterns and Sub-volume PatchMatch Filtering

Visual Computing Lab1
Faculty of Science
University of Ontario Institute of Technology
2000 Simcoe St. N., Oshawa ON L1G 0C5

Best computer vision paper, CRV 2017!

Some results on MPI Sintel PERTURBED_MARKET_3 sequence. Top row (L2R): clean pass, 2 frames/sec (ours EPE 0.99), 4 frames/sec (EPPM-CVPR14 EPE 1.369). Bottom row (L2R): final pass, 2 frames/sec (ours EPE 1.846), 4 frames/sec (EPPM-CVPR14 EPE 2.368).


This paper presents a new method for efficiently computing large-displacement optical flow. The method uses dominant motion patterns to identify a sparse set of sub- volumes within the cost volume and restricts subsequent Edge- Aware Filtering (EAF) to these sub-volumes. The method uses an extension of PatchMatch to filter these sub-volumes. The fact that our method only applies EAF to a small fraction of the entire cost volume boosts runtime performance. We also show that computational complexity is linear in the size of the images and does not depend upon the size of the label space. We evaluate the proposed technique on MPI Sintel, Middlebury and KITTI benchmarks and show that our method achieves accuracy comparable to those of several recent state-of-the-art methods, while posting significantly faster runtimes.


For technical details please look at the following publications