Research map / Video / Video-language understanding
TVHighlights: LLM-Guided Human-Free Collaborative Training for Video Highlight Detection in Movies and TV Dramas
Qi Qiu, Xuan Wu, Jiawei Peng, Yuan Miao, Xu Yang, Yanlong Du
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 Stage 2 LLM-guided iterative refinement (ablation) | LTV-HD Stage 1 · TVHighlights (total) | AP: 62.39 -> 71.20 | Improved |
| add weakly supervised MIL pre-training (ablation) | LTV-HD · TVHighlights (total) | AP: 63.45 -> 67.69 | Improved |
| add weak supervised pre-training before Stage 2 refinement (ablation) | LTV-HD Stage 2 · TVHighlights (total) | AP: 70.01 -> 71.20 | Improved |
| add second refinement round (ablation) | LTV-HD · TVHighlights (total) | AP: 69.78 -> 71.20 | Improved |
| update labels during NLC training (ablation) | NLC learning · TVHighlights (total) | AP: 66.05 -> 70.21 | Improved |
| add compatibility loss (ablation) | NLC learning · TVHighlights (total) | AP: 70.21 -> 70.85 | Improved |
| add sharpness loss (ablation) | NLC learning · TVHighlights (total) | AP: 70.85 -> 71.20 | Improved |
| add sharpness loss (without compatibility loss) (ablation) | NLC learning · TVHighlights (total) | AUC: 93.31 (no cp) vs 92.74 (full) | Mixed |
| use per-clip scoring instead of direct moment retrieval for MLLMs | Qwen-vl-max · TVHighlights (total) | AP: 36.62 -> 47.42 | Improved |
| use NLC noisy-label learning under MTurk label noise | LTV-HD vs UMT · YouTube Highlights (MTurk) | mAP: UMT drops 1.54, ours drops 0.32 | Improved |
| use LTV-HD under clean supervision vs top baseline | LTV-HD · YouTube Highlights (clean) | mAP: 75.8 vs UVCOM2 77.4 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TimeChat — TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Collaborative Noisy Label Cleaner: Learning Scene-Aware Trailers for Multi-Modal Highlight Detection in Movies (CVPR 2023 · similarity 0.91)
- Bridging the Gap: A Unified Video Comprehension Framework for Moment Retrieval and Highlight Detection (CVPR 2024 · similarity 0.86)
- Video Repurposing from User Generated Content: A Large-scale Dataset and Benchmark (AAAI 2025 · similarity 0.85)
- ShotVL: Human-Centric Highlight Frame Retrieval via Language Queries (AAAI 2025 · similarity 0.85)
- Watch Video, Catch Keyword: Context-aware Keyword Attention for Moment Retrieval and Highlight Detection (AAAI 2025 · similarity 0.85)
- Exploring Domain Incremental Video Highlights Detection with the LiveFood Benchmark (AAAI 2024 · similarity 0.84)
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.
