Research map / Vision-language and multimodal / Cross-modal learning
Stepping Stones: A Progressive Training Strategy for Audio-Visual Semantic Segmentation
Juncheng Ma, Peiwen Sun, Yaoting Wang, Di Hu
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 adaptive audio query generator instead of repeated audio features (ablation) | AAVS · AVSBench-Object S4 | mIoU: 80.7 -> 82.5 | Improved |
| add adaptive audio query generator instead of repeated audio features (ablation) | AAVS · AVSBench-Semantic | mIoU: 36.0 -> 37.1 | Improved |
| add mask fusion of first-stage localization with mask feature (ablation) | AAVS with Stepping Stones · AVSBench-Semantic (actual first stage) | mIoU: 45.2 -> 48.5 | Improved |
| add robust audio-aware key generator (ablation) | AAVS with Stepping Stones · AVSBench-Semantic (actual first stage) | mIoU: 46.4 -> 48.5 | Improved |
| use oracle ground-truth first-stage masks instead of predicted ones (ablation) | AAVS with Stepping Stones · AVSBench-Semantic | mIoU: 48.5 -> 64.9 | Improved |
| apply Stepping Stones with low-accuracy first-stage results (ablation) | AVSBench · AVSBench-Semantic | mIoU: 29.8 -> 27.3 | Hurt |
| apply Stepping Stones with high-accuracy first-stage results (ablation) | AVSegformer · AVSBench-Semantic | mIoU: 36.7 -> 39.4 | Improved |
| apply Stepping Stones with high-accuracy first-stage results (ablation) | AVSBench · AVSBench-Semantic | mIoU: 29.8 -> 31.5 | Improved |
Closest papers by meaning
- AVSegFormer: Audio-Visual Segmentation with Transformer (AAAI 2024 · similarity 0.90)
- Prompting Segmentation with Sound Is Generalizable Audio-Visual Source Localizer (AAAI 2024 · similarity 0.90)
- Unraveling Instance Associations: A Closer Look for Audio-Visual Segmentation (CVPR 2024 · similarity 0.90)
- Dynamic Derivation and Elimination: Audio Visual Segmentation with Enhanced Audio Semantics (CVPR 2025 · similarity 0.89)
- Improving Audio-Visual Segmentation with Bidirectional Generation (AAAI 2024 · similarity 0.89)
- Robust Audio-Visual Segmentation via Audio-Guided Visual Convergent Alignment (CVPR 2025 · similarity 0.89)
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.
