Research map / Video / Video-language understanding
Dual Encoding for Video Retrieval by Text
Jianfeng Dong, Xirong Li, Chaoxi Xu, Xun Yang, Gang Yang, Xun Wang, Meng Wang
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 biGRU-CNN level to mean pooling and biGRU (all three levels) (ablation) | Dual Encoding · MSR-VTT official | SumR: 203.8 -> 211.7 | Improved |
| add level 2 (biGRU) to level 1 (mean pooling) (ablation) | Dual Encoding · MSR-VTT official | SumR: 182.9 -> 195.3 | Improved |
| use biGRU-CNN instead of mean pooling as single encoder (ablation) | Dual Encoding · MSR-VTT official | SumR: 182.9 -> 197.9 | Improved |
| add text-side multi-level encoding to video-side multi-level encoding (ablation) | Dual Encoding · MSR-VTT official | SumR: 194.5 -> 211.7 | Improved |
| add video-side multi-level encoding to text-side multi-level encoding (ablation) | Dual Encoding · MSR-VTT official | SumR: 191 -> 211.7 | Improved |
| use hybrid space instead of latent space only (ablation) | Dual Encoding · MSR-VTT official | SumR: 199.0 -> 211.7 | Improved |
| use hybrid space instead of concept space only (ablation) | Dual Encoding · MSR-VTT official | SumR: 187.4 -> 211.7 | Improved |
| increase latent space dimension from 1,536 to 2,048 (ablation) | Dual Encoding · MSR-VTT official | SumR: 199.0 -> 197.7 | No clear effect |
| add triplet ranking loss for concept space learning (ablation) | Dual Encoding · MSR-VTT official | SumR: 183.3 -> 211.7 | Improved |
| add pre-trained Transformer features (MMT, BERT) | Dual Encoding · MSR-VTT Test1k-Yu | text-to-video R@1: 21.1 -> 32.4 | Improved |
| use Transformer features only instead of multi-level encoding (ablation) | Dual Encoding · MSR-VTT official | text-to-video R@1: 11.6 -> 8.5 | Hurt |
| add multi-level text encoding to VSE++ | VSE++ · Flickr30k | SumR: 291.3 -> 310.7 | Improved |
Closest papers by meaning
- Bridging the Semantic Granularity Gap Between Text and Frame Representations for Partially Relevant Video Retrieval (AAAI 2025 · similarity 0.87)
- Towards Efficient and Effective Text-to-Video Retrieval with Coarse-to-Fine Visual Representation Learning (AAAI 2024 · similarity 0.87)
- Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval (CVPR 2025 · similarity 0.87)
- Mitigating Semantic Collapse in Partially Relevant Video Retrieval (NeurIPS 2025 · similarity 0.87)
- MUVR: A Multi-Modal Untrimmed Video Retrieval Benchmark with Multi-Level Visual Correspondence (NeurIPS 2025 · similarity 0.86)
- Text Proxy: Decomposing Retrieval from a 1-to-N Relationship into N 1-to-1 Relationships for Text-Video Retrieval (AAAI 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.
