Research map / Vision-language and multimodal / Visual question answering
VisAssist: A Visually Impaired-Captured Video Question Answering Benchmark for Assistive Systems
Qi Gao, Heng Li, Yixin Zhou, Meixuan Zhou, Jieqiong Chen, Xinyu Chai
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 |
|---|---|---|---|
| sample 1 fps instead of a single middle frame (ablation) | Gemini-2.5-Pro · VisAssist | AVG score: 2.86 -> 3.30 | Improved |
| use FSM keyframes instead of 0.5 fps uniform (ablation) | Gemini-2.5-Pro · VisAssist | AVG score: 3.13 -> 3.15 | Improved |
| use FSM keyframes instead of 0.5 fps uniform (ablation) | Qwen2-VL-OCR · VisAssist | AVG score: 2.05 -> 2.00 | Hurt |
| use FSM keyframes instead of 1 fps (ablation) | Qwen2.5-VL · VisAssist | AVG score / TFLOPs: 2.37 -> 2.19 AVG; 12.64 -> 6.17 TFLOPs | Mixed |
| raise input resolution from 224x224 to 448x448 (ablation) | Gemini-2.5-Pro · VisAssist | AVG score: 3.20 -> 3.30 | Improved |
| raise input resolution from 224x224 to 448x448 (ablation) | VideoLLaMA2 · VisAssist | AVG score: 1.51 -> 1.67 | Improved |
| raise input resolution from 224x224 to 448x448 (ablation) | Qwen2.5-VL · VisAssist | AVG score: 2.34 -> 2.37 | Improved |
| model scale/capability: closed-source models vs open-source video LLMs in zero-shot | VisAssist zero-shot · VisAssist (Chinese) | Avg score: 1.04 (TimeChat) -> 3.30 (Gemini-pro) | Improved |
| fine-tune open-source models on VisAssist training split | Qwen2.5-VL · VisAssist (ZH) | avg COR/DO/SU: 2.37 (zero-shot, answer-only GT) -> 1.865 (fine-tuned, answer+suggestion GT) | Mixed |
Closest papers by meaning
- EgoBlind: Towards Egocentric Visual Assistance for the Blind (NeurIPS 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)
- Explaining CLIP's Performance Disparities on Data from Blind/Low Vision Users (CVPR 2024 · similarity 0.86)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.86)
- MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding (NeurIPS 2024 · similarity 0.86)
- WearVQA: A Visual Question Answering Benchmark for Wearables in Egocentric Authentic Real-world scenarios (NeurIPS 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.
