eduzhai > Applied Sciences > Engineering >

ModeNet Mode Selection Network For Learned Video Coding

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: In this paper, a mode selection network (ModeNet) is proposed to enhance deeplearning-based video compression. Inspired by traditional video coding, ModeNetpurpose is to enable competition among several coding modes. The proposedModeNet learns and conveys a pixel-wise partitioning of the frame, used toassign each pixel to the most suited coding mode. ModeNet is trained alongsidethe different coding modes to minimize a rate-distortion cost. It is a flexiblecomponent which can be generalized to other systems to allow competitionbetween different coding tools. Mod-eNet interest is studied on a P-framecoding task, where it is used to design a method for coding a frame given itsprediction. ModeNet-based systems achieve compelling performance when evaluatedunder the Challenge on Learned Image Compression 2020 (CLIC20) P-frame codingtrack conditions.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...