Research map / Video / Video-language understanding
Language-Guided Graph Representation Learning for Video Summarization
Wenrui Li, Wei Hong Han, Hengyu Man, Wangmeng Zuo, Xiaopeng Fan B, Yonghong Tian
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 BCE (multi-annotation weighted) loss to baseline (ablation) | LGRLN · SumMe (ablation) | F1-max: 47.1 -> 53.9 | Improved |
| add time embedding (Temb) to baseline (ablation) | LGRLN · SumMe (ablation) | F1-max: 47.1 -> 52.2 | Improved |
| add GBT to baseline (ablation) | LGRLN · SumMe (ablation) | F1-max: 47.1 -> 47.7 | No clear effect |
| add GBT to BCE (ablation) | LGRLN · SumMe (ablation) | F1-max: 53.9 -> 54.1 | Improved |
| add remaining module to pairs to form full LGRLN (ablation) | LGRLN · SumMe (ablation) | F1-max: 54.1 -> 55.9 | Improved |
| add each single module to baseline (ablation) | LGRLN · TVSum (ablation) | F1-max: 78.3 -> 80.8 | Mixed |
| add full set of modules (BCE, Temb, GBT) (ablation) | LGRLN · TVSum (ablation) | F1-mean: 56.9 -> 59.5 | Improved |
| increase gap between weights a and b (a=0.000, b=0.500 vs a=0.070, b=0.100) (ablation) | LGRLN · SumMe (weight sweep) | F1-mean: 16.8 -> 22.8 | Mixed |
| use moderate gap between a and b (a=0.020, b=0.160) vs largest gap (ablation) | LGRLN · TVSum (weight sweep) | F1-mean: 42.6 -> 53.9 | Improved |
| use LGRLN instead of LLM-based V2Xum-LLaMA-7B (3M vs 7B parameters) | LGRLN · VideoXum | F1 score: 29.0 -> 32.1 | Improved |
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.
- SumGraph — SumGraph: Video Summarization via Recursive Graph Modeling (ECCV 2020) · better on 3 of 4 shared results
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
- Language-guided Recursive Spatiotemporal Graph Modeling for Video Summarization (IJCV 2025 · similarity 0.93)
- SumGraph: Video Summarization via Recursive Graph Modeling (ECCV 2020 · similarity 0.92)
- Reconstructive Sequence-Graph Network for Video Summarization (TPAMI 2021 · similarity 0.90)
- Video Summarization with Large Language Models (CVPR 2025 · similarity 0.87)
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation (ICML 2025 · similarity 0.87)
- Vgent: Graph-based Retrieval-Reasoning-Augmented Generation For Long Video Understanding (NeurIPS 2025 · similarity 0.86)
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.
