Research map / Video / Video-language understanding
Bridging Video-Text Retrieval With Multiple Choice Questions
Yuying Ge, Yixiao Ge, Xihui Liu, Dian Li, Ying Shan, Xiaohu Qie, Ping Luo
Check 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 MCQ noun- and verb-answering objectives to video-text pre-training (ablation) | dual-encoder video-text pre-training · MSR-VTT, zero-shot | R@1: 22.3 -> 26.0 | Improved |
| add noun and verb questions rather than use randomly erased words (ablation) | MCQ pre-training · MSR-VTT, zero-shot | R@1: 23.0 -> 26.0 | Improved |
| use both noun and verb questions rather than only noun questions (ablation) | MCQ pre-training · MSR-VTT, zero-shot | R@1: 24.9 -> 26.0 | Improved |
| use both noun and verb questions rather than only verb questions (ablation) | MCQ pre-training · MSR-VTT, zero-shot | R@1: 23.3 -> 26.0 | Improved |
| select the answer phrase contrastively rather than predict masked word tokens (ablation) | MCQ pre-training · MSR-VTT, zero-shot | R@1: 20.6 -> 26.0 | Improved |
| use all-level intermediate features rather than only highest-level features (ablation) | BridgeFormer · MSR-VTT, zero-shot | R@1: 23.3 -> 26.0 | Improved |
| use video tokens to answer noun questions (ablation) | BridgeFormer · WebVid-2M validation, noun questions | R@1: 6.6 -> 58.6 | Improved |
| use video tokens to answer verb questions (ablation) | BridgeFormer · WebVid-2M validation, verb questions | R@1: 4.5 -> 40.7 | Improved |
| prepend [MASK] to phrase representations rather than use phrases without a prompt (ablation) | TextFormer phrase representations · MSR-VTT, zero-shot | R@1: 23.1 -> 25.1 | Improved |
| pre-train with MCQ | video-text dual encoder · HowTo100M, 1M videos, zero-shot | R@50: 2.6 -> 3.4 | Improved |
Closest papers by meaning
- MuLTI: Efficient Video-and-Language Understanding with Text-Guided MultiWay-Sampler and Multiple Choice Modeling (AAAI 2024 · similarity 0.88)
- Towards Efficient and Effective Text-to-Video Retrieval with Coarse-to-Fine Visual Representation Learning (AAAI 2024 · similarity 0.88)
- Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval (CVPR 2025 · similarity 0.88)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.87)
- DGL: Dynamic Global-Local Prompt Tuning for Text-Video Retrieval (AAAI 2024 · similarity 0.87)
- MUSE: Mamba Is Efficient Multi-scale Learner 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.
