AI Paper Atlas

Research map / Video / Video segmentation and tracking

Exploring Heterogeneous Clues for Weakly-Supervised Audio-Visual Video Parsing

Yu Wu, Yi Yang

CVPR 2021 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 92 citations (OpenAlex)

Check 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 modality-aware labels from audio/visual track swapping (ablation)Weakly supervised AVVP model · LLP test, VisualSegment-level F-score: 52.9 -> 57.5Improved
Use modality-aware labels from audio/visual track swapping (ablation)Weakly supervised AVVP model · LLP test, Audio-visualSegment-level F-score: 48.9 -> 52.6Improved
Add cross-modal contrastive learning (ablation)Weakly supervised AVVP model · LLP test, Audio-visualSegment-level F-score: 48.9 -> 49.7Improved
Add cross-modal contrastive learning (ablation)Weakly supervised AVVP model · LLP test, AudioSegment-level F-score: 60.1 -> 61.9Improved
Add cross-modal contrastive learning (ablation)Weakly supervised AVVP model · LLP test, VisualSegment-level F-score: 52.9 -> 53.1No clear effect
Add modality-aware label refinement (ablation)Audio parsing model · LLP test, AudioSegment-level F-score: 60.1 -> 59.8Hurt
Add both cross-modal contrastive learning and modality-aware label refinement (ablation)Weakly supervised AVVP model · LLP test, Audio-visualSegment-level F-score: 48.9 -> 55.1Improved
Refine visual labels only rather than audio labels only (ablation)Weakly supervised AVVP model · LLP test, VisualSegment-level F-score: 52.7 -> 59.0Improved
Vary contrastive-learning temperature τ from 0.1 to 0.4 (ablation)Weakly supervised AVVP model · LLP testSegment-level F-score: 58.4; 58.9; 58.7; 58.5No clear effect
Replace HAN with MAAudio-visual parsing · LLP test, Audio-visualSegment-level F-score: 48.9 -> 55.1Improved
Replace HAN with MAAudio-visual parsing · LLP test, AudioSegment-level F-score: 60.1 -> 60.3No clear effect
Replace HAN with MAAudio-visual parsing · LLP test, VisualSegment-level F-score: 52.9 -> 60.0Improved

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