Research map / Video / Video-language understanding
Multi-Event Video-Text Retrieval
Gengyuan Zhang, Jisen Ren, Jindong Gu, Volker Tresp
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 key event selection (K-Medoids frames) (ablation) | Me-Retriever · ActivityNet Captions Video-to-Text | Recall@1-Average: 7.40 -> 8.52 | Improved |
| add MeVTR loss instead of cross-entropy (ablation) | Me-Retriever · ActivityNet Captions Video-to-Text | Recall@1-Average: 6.70 -> 8.52 | Improved |
| add key event selection (ablation) | Me-Retriever · ActivityNet Captions Text-to-Video | Recall@1: 15.03 -> 15.17 | Improved |
| add MeVTR loss instead of cross-entropy (ablation) | Me-Retriever · ActivityNet Captions Text-to-Video | Recall@1: 14.45 -> 15.17 | Improved |
| use dynamic weight alpha instead of fixed alpha (ablation) | MeVTR loss · ActivityNet Captions Video-to-Text and Text-to-Video | Recall@5 | Improved |
| increase fixed alpha beyond 1.0-2.0 (ablation) | MeVTR loss · ActivityNet Captions | Recall@5 | Mixed |
| use max instead of avg key-event similarity (ablation) | Me-Retriever · Charades-Event Text-to-Video | Recall@1: 1.88 -> 2.02 | Improved |
| use max instead of avg key-event similarity (ablation) | Me-Retriever · ActivityNet Captions Text-to-Video | Recall@1: 15.99 -> 14.93 | Hurt |
| use max instead of avg key-event similarity (ablation) | Me-Retriever · ActivityNet Captions Video-to-Text | Recall@1-Average: 8.52 -> 6.98 | Hurt |
| use Me-Retriever(avg) instead of CLIP4Clip(mean) (ablation) | video encoder aggregation · ActivityNet Captions Video-to-Text | Recall@1-Average: 6.60 -> 8.52 | Improved |
| use Me-Retriever(avg) instead of CLIP4Clip(mean) (ablation) | video encoder aggregation · ActivityNet Captions Text-to-Video | Recall@1: 15.25 -> 15.99 | No clear effect |
| use Me-Retriever(avg) instead of CLIP4Clip(mean) | video encoder aggregation · Charades-Event Video-to-Text | Recall@1-Average: 1.32 -> 1.39 | Improved |
Closest papers by meaning
- MultiVENT 2.0: A Massive Multilingual Benchmark for Event-Centric Video Retrieval (CVPR 2025 · similarity 0.88)
- MV-Adapter: Multimodal Video Transfer Learning for Video Text Retrieval (CVPR 2024 · similarity 0.88)
- MUVR: A Multi-Modal Untrimmed Video Retrieval Benchmark with Multi-Level Visual Correspondence (NeurIPS 2025 · similarity 0.88)
- Text Proxy: Decomposing Retrieval from a 1-to-N Relationship into N 1-to-1 Relationships for Text-Video Retrieval (AAAI 2025 · similarity 0.87)
- Bridging Information Asymmetry in Text-video Retrieval: A Data-centric Approach (ICLR 2025 · similarity 0.87)
- Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval (CVPR 2025 · similarity 0.87)
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.
