Research map / Video / Video-language understanding
Multi-Speaker Video Dialog with Frame-Level Temporal Localization
Qiang Wang, Pin Jiang, Zhiyi Guo, Yahong Han, Zhou Zhao
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 chat history plus video instead of chat history only (ablation) | SelRef · MSVD w/o TL (20s) | R@1: 17.9 -> 21.4 | Improved |
| use chat history plus video instead of video only (ablation) | SelRef · MSVD w/o TL (20s) | R@1: 19.5 -> 21.4 | Improved |
| use chat history plus video instead of chat history only (ablation) | SelRef · MSVD-TL (50s) | R@1: 17.3 -> 21.5 | Improved |
| use chat history plus video instead of video only (ablation) | SelRef · MSVD-TL (50s) | R@1: 19.5 -> 21.5 | Improved |
| SelRef relation modeling vs Triple Encoder | SelRef (C+V) · MSVD w/o TL (20s) | R@1: 18.1 -> 21.4 | Improved |
| SelRef vs TriDAF+self Attn baseline | SelRef (C+V) · MSVD w/o TL (20s) | R@1: 20.7 -> 21.4 | Improved |
| extend input video from 20s sub-segment to 50s fragment without temporal localization | SelRef (C+V) · MSVD w/o TL | R@5: 69.7 -> 69.1 | Hurt |
| extend input video from 20s to 50s without temporal localization | TriDAF+self Attn (C+V) · MSVD w/o TL | R@5: 69.4 -> 68.2 | Hurt |
| add frame-level temporal localization on 50s fragment | SelRef (C+V) · 50s fragment | R@1: 20.9 -> 21.5 | Improved |
| add temporal localization on 50s fragment | SelRef (C+V) · 50s fragment | R@5: 69.1 -> 70.1 | Improved |
| add temporal localization on 50s fragment | SelRef (V) · 50s fragment | R@1: 20.2 -> 19.5 | No clear effect |
| add temporal localization on 50s fragment | SelRef (C) · 50s fragment | R@1: 17.4 -> 17.3 | No clear effect |
Closest papers by meaning
- V^2Dial: Unification of Video and Visual Dialog via Multimodal Experts (CVPR 2025 · similarity 0.87)
- Friends-MMC: A Dataset for Multi-modal Multi-party Conversation Understanding (AAAI 2025 · similarity 0.86)
- SAMA: Towards Multi-Turn Referential Grounded Video Chat with Large Language Models (NeurIPS 2025 · similarity 0.86)
- Structure-Aware Multimodal Sequential Learning for Visual Dialog (AAAI 2024 · similarity 0.85)
- MMDU: A Multi-Turn Multi-Image Dialog Understanding Benchmark and Instruction-Tuning Dataset for LVLMs (NeurIPS 2024 · similarity 0.85)
- Structured Video-Language Modeling with Temporal Grouping and Spatial Grounding (ICLR 2024 · similarity 0.85)
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.
