AI Paper Atlas

Research map / Video / Video segmentation and tracking

Hierarchical B-Frame Video Coding Using Two-Layer CANF Without Motion Coding

David Alexandre, Hsueh-Ming Hang, Wen-Hsiao Peng

CVPR 2023 · Video › Video segmentation and tracking · approach: Learned video compression · 23 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use a two-layer low-resolution base and full-resolution enhancement codec without transmitting motion informationTLZMC · UVG and HEVC Class BPSNRImproved
Use a two-layer codec without motion coding instead of B-CANFTLZMC · UVG and HEVC Class BMS-SSIMHurt
Add frame-type adaptive coding (ablation)TLZMC · UVG and HEVC Class Brate-distortion performanceImproved
Add a low-resolution base layer to convey unpredictable-pixel informationTLZMC · B-frame video codingrate-distortion performanceImproved
Remove skip-mask coding (ablation)TLZMC · UVG, GOP=32BD-rate saving: -42.39 vs -35.87Hurt
Add predicted motion fields and merged image as skip-mask inputs (ablation)TLZMC skip-mask generator · UVG, GOP=32BD-rate saving: -35.87 -> -38.12Improved
Add hyperprior μ and σ as skip-mask inputs (ablation)TLZMC skip-mask generator · UVG, GOP=32BD-rate saving: -35.87 -> -39.74Improved
Combine motion/merged-image clues with hyperprior clues for skip-mask generation (ablation)TLZMC skip-mask generator · UVG, GOP=32BD-rate saving: -38.12 and -39.74 -> -42.39Improved
Increase λ, allowing more latent samples to be retainedTLZMC MSE model · UVGretained latent samples: 28.28%, 36.23%, 57.17%, and 69.93%Improved
Use GOP=32 rather than GOP=8 (ablation)TLZMC · UVGBD-rate savingImproved
Replace the multi-frame merging network with frame synthesis (ablation)TLZMC · B-frame video codingrate-distortion performanceImproved
Use TLZMC instead of B-CANFB-frame video coding · UVG, GOP=32encode MACs: 2.70M/px -> 1.50M/pxImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy