Research map / Vision-language and multimodal / Visual question answering
Knowledge Acquisition for Visual Question Answering via Iterative Querying
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 a learned query generator instead of the MLP baseline without acquired evidence | MLP [17] · Visual7W telling test | accuracy: 0.648 -> 0.679 | Helped |
| Use a learned query generator instead of querying uniformly at random | MLP + iterative querying · Visual7W telling test | accuracy: 0.653 -> 0.679 | Helped |
| Use a learned query generator instead of putting all scene-graph knowledge in memory | MLP + iterative querying · Visual7W telling test | accuracy: 0.658 -> 0.679 | Helped |
| Use a learned query generator instead of the MLP baseline without acquired evidence | MLP [17] · VQA test-dev | all: 0.659 -> 0.674 | Helped |
| Use a learned query generator instead of querying uniformly at random | MLP + iterative querying · VQA test-dev | all: 0.666 -> 0.674 | Helped |
| Use a learned query generator instead of putting all detected objects in memory | MLP + iterative querying · VQA test-dev | all: 0.668 -> 0.674 | Helped |
| Use a learned query generator instead of the MLP baseline without acquired evidence | MLP [17], trained on train+val · VQA test-dev | all: 0.659 -> 0.691 | Helped |
| Add object-detection evidence through iterative queries | MLP [17], trained on train+val · VQA test-standard | all | Helped |
| Train the query-generator model on train+val rather than train (ablation) | MLP + query generator · VQA test-dev | all: 0.674 -> 0.691 | Helped |
Closest papers by meaning
- TOA: Task-oriented Active VQA (NeurIPS 2023 · similarity 0.92)
- Prompting Large Language Models With Answer Heuristics for Knowledge-Based Visual Question Answering (CVPR 2023 · similarity 0.92)
- Learning by Asking Questions (CVPR 2018 · similarity 0.91)
- OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge (CVPR 2019 · similarity 0.91)
- Separation of Powers: On Segregating Knowledge from Observation in LLM-enabled Knowledge-based Visual Question Answering (CVPR 2025 · similarity 0.90)
- SkillCLIP: Skill Aware Modality Fusion Visual Question Answering (Student Abstract) (AAAI 2024 · similarity 0.90)
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.