Research map / Vision-language and multimodal / Cross-modal learning
Whence the Voice? Self-supervised Dual-source Audio-Visual Localisation via Selective Convergence
Han Hu, Dongheng Lin, Yuqi Hou, Haotian LI, Hyung Jin Chang, Jianbo Jiao
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 |
|---|---|---|---|
| Use learned spatial prior from Stage 1 instead of uniform mask (ablation) | SCAV · VGGSound-Duet (BBox) ablation | CIoU@0.3 (%): 28.94 -> 53.27 | Improved |
| Progressive two-stage training instead of joint training (ablation) | SCAV · VGGSound-Duet (BBox) ablation | CIoU@0.3 (%): 41.84 -> 53.27 | Improved |
| Exclude cross-sample negatives in Stage 2 (ablation) | SCAV · VGGSound-Duet (BBox) ablation | CIoU@0.3 (%): 26.69 -> 53.27 | Improved |
| Add dominant-source post-processing (ablation) | SCAV · VGGSound-Duet (BBox) ablation | CIoU@0.3 (%): 52.44 -> 53.27 | No clear effect |
| Progressive training on mask-based benchmark (ablation) | SCAV · VGGSound-Duet Mask ablation | CIoU@0.3 (%): 25.62 -> 33.59 | Improved |
| Stage 1 contrastive training on mixed audio | Stage 1 model · VGGSound-Duet, first-stage model | CIoU@0.3 (%): dominant 61.70 vs second 3.62 | Mixed |
| SCAV vs best self-supervised baseline on mask evaluation | SCAV · VGGSound-Duet Mask | CIoU@0.3 (%): 9.58 -> 33.59 | Improved |
| SCAV vs AVGN on MUSIC-Duet detection | SCAV · MUSIC-Duet (BBox) | CAP (%): 50.6 -> 47.6 | Hurt |
| SCAV vs OA-SSL on VGGSound-Duet bbox | SCAV · VGGSound-Duet (BBox, frame-wise) | CIoU@0.3 (%): 55.2 -> 53.3 | Hurt |
Closest papers by meaning
- Robust Audio-Visual Contrastive Learning for Proposal-Based Self-Supervised Sound Source Localization in Videos (TPAMI 2024 · similarity 0.91)
- Prompting Segmentation with Sound Is Generalizable Audio-Visual Source Localizer (AAAI 2024 · similarity 0.89)
- Seeing Speech and Sound: Distinguishing and Locating Audio Sources in Visual Scenes (CVPR 2025 · similarity 0.89)
- T-VSL: Text-Guided Visual Sound Source Localization in Mixtures (CVPR 2024 · similarity 0.89)
- Object-aware Sound Source Localization via Audio-Visual Scene Understanding (CVPR 2025 · similarity 0.89)
- Robust Audio-Visual Segmentation via Audio-Guided Visual Convergent Alignment (CVPR 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.
