eduzhai > Applied Sciences > Engineering >

Chroma Intra Prediction with attention-based CNN architectures

  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Neural networks can be used in video coding to improve chromaintra-prediction. In particular, usage of fully-connected networks has enabledbetter cross-component prediction with respect to traditional linear models.Nonetheless, state-of-the-art architectures tend to disregard the location ofindividual reference samples in the prediction process. This paper proposes anew neural network architecture for cross-component intra-prediction. Thenetwork uses a novel attention module to model spatial relations betweenreference and predicted samples. The proposed approach is integrated into theVersatile Video Coding (VVC) prediction pipeline. Experimental resultsdemonstrate compression gains over the latest VVC anchor compared withstate-of-the-art chroma intra-prediction methods based on neural networks.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×