Research map / Video / Video-language understanding
Dual Encoding for Zero-Example Video Retrieval
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use dual multi-level encoding for both video and text | Video retrieval · MSR-VTT text-to-video | sum of recalls: 132.1 -> 148.6 | Helped |
| Add Level 3 biGRU-CNN to Level 1 and Level 2 (ablation) | Dual encoding · MSR-VTT | sum of recalls: 147.6 -> 148.6 | Helped |
| Add Level 2 biGRU to Level 1 mean pooling (ablation) | Dual encoding · MSR-VTT | sum of recalls: 124.4 -> 138.3 | Helped |
| Add Level 3 biGRU-CNN to Level 1 mean pooling (ablation) | Dual encoding · MSR-VTT | sum of recalls: 124.4 -> 144.7 | Helped |
| Add Level 3 biGRU-CNN to Level 2 biGRU (ablation) | Dual encoding · MSR-VTT | sum of recalls: 124.8 -> 147.6 | Helped |
| Use Level 3 biGRU-CNN alone instead of Level 1 mean pooling (ablation) | Video and text encoding · MSR-VTT | sum of recalls: 124.4 -> 136.4 | Helped |
| Use multi-level encoding on both sides instead of video-side encoding with text mean pooling (ablation) | Dual encoding · MSR-VTT | sum of recalls: 143.6 -> 148.6 | Helped |
| Use multi-level encoding on both sides instead of text-side encoding with video mean pooling (ablation) | Dual encoding · MSR-VTT | sum of recalls: 137.1 -> 148.6 | Helped |
| Train with MSR-VTT and TGIF together rather than either dataset alone (ablation) | Dual encoding · TRECVID 2016 AVS | infAP: 0.140 -> 0.159; 0.092 -> 0.159 | Helped |
| Train with MSR-VTT and TGIF together rather than either dataset alone (ablation) | Dual encoding · TRECVID 2017 AVS | infAP: 0.179 -> 0.208; 0.199 -> 0.208 | Helped |
| Replace VSE++ GRU text encoding with multi-level text encoding (ablation) | VSE++ · Flickr30k text-to-image | R@1: 23.1 -> 24.7 | Helped |
| Replace VSE++ GRU text encoding with multi-level text encoding (ablation) | VSE++ · MSCOCO text-to-image | R@1: 33.7 -> 34.8 | Helped |
Closest papers by meaning
- MUVR: A Multi-Modal Untrimmed Video Retrieval Benchmark with Multi-Level Visual Correspondence (NeurIPS 2025 · similarity 0.85)
- How to Make Cross Encoder a Good Teacher for Efficient Image-Text Retrieval? (CVPR 2024 · similarity 0.84)
- DGL: Dynamic Global-Local Prompt Tuning for Text-Video Retrieval (AAAI 2024 · similarity 0.83)
- Towards Balanced Alignment: Modal-Enhanced Semantic Modeling for Video Moment Retrieval (AAAI 2024 · similarity 0.83)
- Bridging the Semantic Granularity Gap Between Text and Frame Representations for Partially Relevant Video Retrieval (AAAI 2025 · similarity 0.83)
- Mitigating Semantic Collapse in Partially Relevant Video Retrieval (NeurIPS 2025 · similarity 0.83)
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.