no code implementations • 22 Jan 2019 • Ijaz Akhter, Cheong Loong Fah, Richard Hartley
We propose a new video representation in terms of an over-segmentation of dense trajectories covering the whole video.
Position Superpixels