Research map / Video / Video-language understanding
Simple Baselines for Interactive Video Retrieval with Questions and Answers
Kaiqu Liang, Samuel Albanie
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 Heuristic question-answer interaction to BLIP | BLIP · MSR-VTT 1k-A | R@1: 43.3 -> 67.9 | Improved |
| add Heuristic question-answer interaction to BLIP | BLIP · MSVD | R@1: 44.2 -> 74.2 | Improved |
| add Heuristic question-answer interaction to BLIP | BLIP · AVSD | R@1: 27.3 -> 39.5 | Improved |
| add Ask Object (AO) augmentation to Auto-text (ablation) | Auto-text question generator · MSR-VTT 1k-A | R@1: 47.6 -> 62.2 | Improved |
| add Ask Object (AO) augmentation to Auto-text-vid (ablation) | Auto-text-vid question generator · MSR-VTT 1k-A | R@1: 49.0 -> 57.6 | Improved |
| condition question generation on top-k retrieved video captions (Auto-text-vid vs Auto-text), with AO (ablation) | Auto-text · MSR-VTT 1k-A | R@1: 62.2 -> 57.6 | Hurt |
| condition question generation on top-k retrieved video captions (Auto-text-vid vs Auto-text), without AO (ablation) | Auto-text · MSR-VTT 1k-A | R@1: 47.6 -> 49.0 | Improved |
| answer with VideoQA model instead of caption plus language model (CAP+LM) (ablation) | Interactive retrieval answer generator · 50 MSR-VTT videos, Heuristic questions | R@1: 60.0 -> 66.0 | Improved |
| answer with VideoQA model instead of caption plus language model (CAP+LM) (ablation) | Interactive retrieval answer generator · 50 MSR-VTT videos | avg answer time (s): 8.0 -> 0.1 | Improved |
| simulate user with VideoQA instead of human answers (ablation) | Heuristic question generator · 50 MSR-VTT videos | R@1: 69.0 (human) vs 66.0 (VideoQA) | No clear effect |
| concatenate all 10 rounds of ground-truth AVSD dialog with the query | BLIP · AVSD | R@1: 27.3 -> 27.4 | No clear effect |
| add Heuristic interaction to BLIP on long-form video | BLIP · ActivityNet | R@1: 35.3 -> 44.8 | Improved |
Closest papers by meaning
- Bridging Video-Text Retrieval With Multiple Choice Questions (CVPR 2022 · similarity 0.89)
- Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval (CVPR 2025 · similarity 0.88)
- VideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinement (ECCV 2026 · similarity 0.88)
- Streaming Video Question-Answering with In-context Video KV-Cache Retrieval (ICLR 2025 · similarity 0.87)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · 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.
