Research map / Vision-language and multimodal / Cross-modal learning
Unified Multisensory Perception: Weakly-Supervised Audio-Visual Video Parsing
Yapeng Tian, Dingzeyu Li, Chenliang Xu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add guided loss L_g to weak loss L_wsl (ablation) | MMIL model without HAN · LLP Visual segment-level | F-score: 16.4 -> 51.7 | Improved |
| add guided loss L_g to weak loss L_wsl (ablation) | MMIL model without HAN · LLP Audio segment-level | F-score: 56.9 -> 45.1 | Hurt |
| add L_wsl to L_g only (ablation) | MMIL model without HAN · LLP Type@AV segment-level | F-score: 40.3 -> 44.0 | Improved |
| use attentive MMIL pooling instead of max pooling (ablation) | MMIL model without HAN · LLP Type@AV segment-level | F-score: 32.6 -> 44.0 | Improved |
| use attentive MMIL pooling instead of mean pooling (ablation) | MMIL model without HAN · LLP Type@AV segment-level | F-score: 39.5 -> 44.0 | Improved |
| use attentive MMIL pooling instead of max pooling (ablation) | HAN model · LLP Type@AV segment-level | F-score: 52.1 -> 53.2 | Improved |
| add GRU temporal network (ablation) | MMIL model · LLP Type@AV segment-level | F-score: 44.0 -> 46.8 | Improved |
| add Transformer temporal network (ablation) | MMIL model · LLP Type@AV segment-level | F-score: 44.0 -> 49.7 | Improved |
| add hybrid attention network (HAN) (ablation) | MMIL model · LLP Type@AV segment-level | F-score: 44.0 -> 53.2 | Improved |
| add L_wsl to L_g only (ablation) | HAN model · LLP Type@AV segment-level | F-score: 52.5 -> 53.2 | Improved |
| add guided loss L_g to L_wsl (ablation) | HAN model · LLP Audio-Visual segment-level | F-score: 20.1 -> 48.4 | Improved |
| add Bootstrap noisy-label handling (ablation) | HAN model · LLP Type@AV segment-level | F-score: 53.2 -> 53.1 | No clear effect |
Closest papers by meaning
- Weakly-Supervised Audio-Visual Video Parsing with Prototype-based Pseudo-Labeling (CVPR 2024 · similarity 0.91)
- UWAV: Uncertainty-weighted Weakly-supervised Audio-Visual Video Parsing (CVPR 2025 · similarity 0.90)
- Multimodal Class-aware Semantic Enhancement Network for Audio-Visual Video Parsing (AAAI 2025 · similarity 0.89)
- Adapting to the Unknown: Training-Free Audio-Visual Event Perception with Dynamic Thresholds (CVPR 2025 · similarity 0.89)
- PreFM: Online Audio-Visual Event Parsing via Predictive Future Modeling (NeurIPS 2025 · similarity 0.88)
- Dense Audio-Visual Event Localization Under Cross-Modal Consistency and Multi-Temporal Granularity Collaboration (AAAI 2025 · similarity 0.88)
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.
