Research map / Vision-language and multimodal / Cross-modal learning
TAViS: Text-bridged Audio-Visual Segmentation with Foundation Models
Ziyang Luo, Nian Liu, Xuguang Yang, Salman Khan, Rao Muhammad Anwer, Hisham Cholakkal, Fahad Shahbaz Khan, Junwei Han
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 ImageBind-guided query decomposition (ablation) | TAViS · MS3 | mIoU (M_J): 58.5 -> 68.2 | Improved |
| add ImageBind-guided query decomposition (ablation) | TAViS · S4 | mIoU (M_J): 79.6 -> 84.8 | Improved |
| add text-bridged alignment supervision (ablation) | TAViS · MS3 | mIoU (M_J): 64.9 -> 68.2 | Improved |
| add text-bridged hybrid prompting (ablation) | TAViS · MS3 | mIoU (M_J): 65.1 -> 68.2 | Improved |
| add audio-to-text loss (ablation) | TAViS · MS3 | mIoU (M_J): 65.2 -> 68.2 | Improved |
| add image-to-text loss (ablation) | TAViS · MS3 | mIoU (M_J): 64.4 -> 68.2 | Improved |
| add direct audio-image alignment loss L_a2i (ablation) | TAViS · MS3 | mIoU (M_J): 68.2 -> 66.5 | Hurt |
| use only L_a2i instead of text-bridged losses (ablation) | TAViS · MS3 | mIoU (M_J): 68.2 -> 66.5 | Hurt |
| add pseudo-text sparse prompt p^t (ablation) | TAViS · MS3 | mIoU (M_J): 63.4 -> 68.2 | Improved |
| use ImageBind text encoder instead of MLP prediction of pseudo-text (ablation) | TAViS · MS3 | mIoU (M_J): 63.6 -> 68.2 | Improved |
| add image cls dense prompt t_v (ablation) | TAViS · MS3 | mIoU (M_J): 63.6 -> 68.2 | Improved |
| increase image size from 224 to 1024 | TAViS · MS3 | mIoU (M_J): 68.2 -> 71.2 | Improved |
Closest papers by meaning
- Unraveling Instance Associations: A Closer Look for Audio-Visual Segmentation (CVPR 2024 · similarity 0.91)
- Unsupervised Audio-Visual Segmentation with Modality Alignment (AAAI 2025 · similarity 0.90)
- Robust Audio-Visual Segmentation via Audio-Guided Visual Convergent Alignment (CVPR 2025 · similarity 0.89)
- TSAM: Temporal SAM Augmented with Multimodal Prompts for Referring Audio-Visual Segmentation (CVPR 2025 · similarity 0.88)
- SAM2-LOVE: Segment Anything Model 2 in Language-aided Audio-Visual Scenes (CVPR 2025 · similarity 0.88)
- Dynamic Derivation and Elimination: Audio Visual Segmentation with Enhanced Audio Semantics (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.
