Research map / Vision-language and multimodal / Cross-modal learning
Watch, Listen and Tell: Multi-Modal Weakly Supervised Dense Event Captioning
Tanzila Rahman, Bicheng Xu, Leonid Sigal
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 MFCC audio features to C3D video features (MUTAN fusion) | unimodal C3D captioner · ActivityNet Captions final model | BLEU@4: 0.83 -> 0.90 | Improved |
| add MFCC audio features to C3D video features (MUTAN fusion) | unimodal C3D captioner · ActivityNet Captions final model | mIoU: 29.78 -> 30.02 | Improved |
| add MFCC audio features to C3D video features | unimodal C3D captioner · ActivityNet Captions final model | CIDEr: 13.81 -> 13.79 | No clear effect |
| add SoundNet audio features to C3D video features | unimodal C3D captioner · ActivityNet Captions final model | METEOR: 4.89 -> 5.03 | Improved |
| add MFCC audio to video features | unimodal C3D captioner · ActivityNet Captions (ground-truth segments) | CIDEr: 24.46 -> 25.36 | Improved |
| add SoundNet audio to video features | unimodal C3D captioner · ActivityNet Captions (ground-truth segments) | CIDEr: 24.46 -> 24.22 | Hurt |
| use audio only instead of video only | unimodal captioner · ActivityNet Captions final model | CIDEr: 13.81 -> 9.56 | Hurt |
| use MUTAN fusion instead of multiplicative mixture fusion (ablation) | MFCC + C3D multi-modal model · ActivityNet Captions final model | mIoU: 29.87 -> 30.02 | Improved |
| use MUTAN fusion instead of multi-modal context fusion (ablation) | MFCC + C3D multi-modal model · ActivityNet Captions pretrained model | CIDEr: 7.91 -> 8.20 | Improved |
| use MUTAN fusion instead of multiplicative mixture fusion (ablation) | MFCC + C3D multi-modal model · ActivityNet Captions final model | CIDEr: 13.97 -> 13.79 | Hurt |
| use MFCC audio instead of CQT audio | audio-only captioner · ActivityNet Captions final model | CIDEr: 8.97 -> 9.56 | Improved |
| add multi-modal MFCC audio with zero vectors for missing audio | pretrained captioner · ActivityNet Captions full dataset | CIDEr: 10.45 -> 10.53 | Improved |
Closest papers by meaning
- CLASP: Cross-modal Salient Anchor-based Semantic Propagation for Weakly-supervised Dense Audio-Visual Event Localization (AAAI 2026 · similarity 0.91)
- Exploring Heterogeneous Clues for Weakly-Supervised Audio-Visual Video Parsing (CVPR 2021 · similarity 0.91)
- Dense Audio-Visual Event Localization Under Cross-Modal Consistency and Multi-Temporal Granularity Collaboration (AAAI 2025 · similarity 0.91)
- LongVALE: Vision-Audio-Language-Event Benchmark Towards Time-Aware Omni-Modal Perception of Long Videos (CVPR 2025 · similarity 0.90)
- Audio-Visual Semantic Graph Network for Audio-Visual Event Localization (CVPR 2025 · similarity 0.89)
- Empowering LLMs with Pseudo-Untrimmed Videos for Audio-Visual Temporal Understanding (AAAI 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.
