eduzhai > Applied Sciences > Engineering >

Optical Flow and Mode Selection for Learning-based Video Coding

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: This paper introduces a new method for inter-frame coding based on twocomplementary autoencoders: MOFNet and CodecNet. MOFNet aims at computing andconveying the Optical Flow and a pixel-wise coding Mode selection. The opticalflow is used to perform a prediction of the frame to code. The coding modeselection enables competition between direct copy of the prediction ortransmission through CodecNet. The proposed coding scheme is assessed under theChallenge on Learned Image Compression 2020 (CLIC20) P-frame coding conditions,where it is shown to perform on par with the state-of-the-art video codecITU MPEG HEVC. Moreover, the possibility of copying the prediction enables tolearn the optical flow in an end-to-end fashion i.e. without relying onpre-training and or a dedicated loss term.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...