Optical Flow Estimation with Deep Learning, a Survey on Recent Advances

One of the many components used in biometrics is optical flow estimation. This could be due to the fact that motion is an inseparable attribute of our (visual) world and hence it is a valuable resource of data needed to tackle many real-world problems. Indeed, technologies that use object detection, motion detection, object tracking, gait recognition as well as video compression heavily rely on optical flow estimation. This chapter explores recent advances in optical flow estimation, while mainly focusing on estimation techniques based on deep learning (DL). In fact, recent advancements in deep learning are seemingly making a shift in the optical flow estimation research field. This chapter begins with reviewing traditional (handcrafted) approaches, then introduces the more recent approaches, and finally gets concluded with surveying deep learning approaches.