Research map / Video / Video-language understanding
Unified Coarse-to-Fine Alignment for Video-Text Retrieval
Ziyang Wang, Yi-Lin Sung, Feng Cheng, Gedas Bertasius, Mohit Bansal
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 frame-sentence alignment to video-sentence (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 44.5 -> 47.1 | Improved |
| add patch-word alignment (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 47.1 -> 48.2 | Improved |
| add patch-sentence alignment on top of patch-word (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 48.2 -> 47.2 | Hurt |
| use ISA instead of mean pooling for frame-sentence aggregation (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 45.6 -> 47.1 | Improved |
| use ISA instead of softmax weight for frame-sentence aggregation (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 46.2 -> 47.1 | Improved |
| use bidirectional Bi-ISA instead of direct ISA for patch-word matrix (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 46.4 -> 48.2 | Improved |
| add Sinkhorn-Knopp normalization of each level before summing (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@1: 48.2 -> 49.4 | Improved |
| add SK norm (ablation) | UCoFiA · MSR-VTT (text-to-video) | R@5: 73.3 -> 72.1 | Hurt |
| UCoFiA vs X-CLIP on MSR-VTT text-to-video | UCoFiA · MSR-VTT (text-to-video) | R@1: 46.1 -> 49.4 | Improved |
| add SK normalization on multi-caption MSVD (ablation) | UCoFiA · MSVD | R@1 | Improved |
| UCoFiA vs TS2-Net on MSR-VTT R@5 | UCoFiA · MSR-VTT (text-to-video) | R@5: 74.5 -> 72.1 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- TS2-Net — TS2-Net: Token Shift and Selection Transformer for Text-Video Retrieval (ECCV 2022) · better on 2 of 3 shared results
- X-Pool — X-Pool: Cross-Modal Language-Video Attention for Text-Video Retrieval (CVPR 2022) · 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
- Towards Efficient and Effective Text-to-Video Retrieval with Coarse-to-Fine Visual Representation Learning (AAAI 2024 · similarity 0.90)
- DiscoVLA: Discrepancy Reduction in Vision, Language, and Alignment for Parameter-Efficient Video-Text Retrieval (CVPR 2025 · similarity 0.89)
- Mitigating Semantic Collapse in Partially Relevant Video Retrieval (NeurIPS 2025 · similarity 0.88)
- Bridging the Semantic Granularity Gap Between Text and Frame Representations for Partially Relevant Video Retrieval (AAAI 2025 · similarity 0.88)
- Bridging Information Asymmetry in Text-video Retrieval: A Data-centric Approach (ICLR 2025 · similarity 0.87)
- 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)
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.
