Publication year and journal Link
Introduction
- Optical flow is the task of estimating the per-pixel motion between video frames. If the camera moves during video then, it also occurs against background.

- Mainly of two types i.e sparse optical flow estimation and dense optical flow estimation.
- Sparse optical flow estimation: Optical flow was done only for some specific regions which are highly characterized such as corners and boundaries
- Dense optical flow estimation: it is done for every pixel. Mainly the deep learning approach is employed here.

- The color representation defines the specific direction of movements for sparse optical flow

Contribution
- Supervised deep learning based methods where the features and motion priors are learned unlike traditional approaches which uses the hand-crafted methods.
- It operates on high resolution flow field thereby apply to small fast moving objects also.