ScopeFlow: Dynamic Scene Scoping for Optical Flow
This work addresses a key training bottleneck in optical flow estimation for computer vision applications, offering significant performance boosts with minimal overhead.
The paper tackles the problem of bias in sampling challenging data during optical flow training, proposing a modified training protocol that improves accuracy without increasing computational complexity. The method achieves over 10% improvement on the MPI Sintel benchmark and sets new state-of-the-art results on KITTI benchmarks, with up to 19.7% gains.
We propose to modify the common training protocols of optical flow, leading to sizable accuracy improvements without adding to the computational complexity of the training process. The improvement is based on observing the bias in sampling challenging data that exists in the current training protocol, and improving the sampling process. In addition, we find that both regularization and augmentation should decrease during the training protocol. Using an existing low parameters architecture, the method is ranked first on the MPI Sintel benchmark among all other methods, improving the best two frames method accuracy by more than 10%. The method also surpasses all similar architecture variants by more than 12% and 19.7% on the KITTI benchmarks, achieving the lowest Average End-Point Error on KITTI2012 among two-frame methods, without using extra datasets.