AI Paper Atlas

Research map / Video / Video-language understanding

Evidence-Backed Video Question Answering

Shijie Wang, Honglu Zhou, Ziyang Wang, Ran Xu, Caiming Xiong, Silvio Savarese, Chen Sun, Juan Carlos Niebles

ECCV 2026 · Video › Video-language understanding · approach: Video question answering · 3 citations (OpenAlex)

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 changedWhereMeasuredEffect
fine-tune UniPixel-7B on ST-Evidence-InstructUniPixel-7B · ST-Evidence-Gent-mean: 1.9 -> 29.1Improved
fine-tune UniPixel-7B on ST-Evidence-InstructUniPixel-7B · ST-Evidence-GenJ&F: 40.3 -> 54.1Improved
fine-tune UniPixel-3B on ST-Evidence-InstructUniPixel-3B · ST-Evidence-Gent-mean: 6.5 -> 26.9Improved
fine-tune UniPixel-3B on ST-Evidence-InstructUniPixel-3B · ST-Evidence-GenQA accuracy: 72.2 -> 72.4No clear effect
fine-tune UniPixel-7B on ST-Evidence-InstructUniPixel-7B · MeViS_uJ&F: 61.7 -> 62.5Improved
fine-tune UniPixel-3B on ST-Evidence-InstructUniPixel-3B · MVBenchaccuracy: 62.5 -> 62.3No clear effect
corrupt annotated evidence regions (ablation)Gemini-2.5-Flash · ST-Evidence-Gen (200-sample subset)QA accuracy: 81.32 -> 61.12Hurt
corrupt non-evidence regions (ablation)Gemini-2.5-Flash · ST-Evidence-Gen (200-sample subset)QA accuracy: 81.32 -> 76.43Hurt
corrupt annotated evidence regions (ablation)Qwen3-VL-8B · ST-Evidence-Gen (200-sample subset)QA accuracy: 79.31 -> 58.32Hurt
replace Qwen2.5-VL-72B with the smaller Qwen3-VL-4BQwen-VL family · ST-Evidence-Gent-mean: 25.3 -> 38.6Improved
scale model size within a familyQwen2.5-VL / Qwen3-VL · ST-Evidence-GenQA accuracy vs evidenceMixed
use manual referring expressions instead of Gemini-2.5-Pro expressions with proxy segmenter (ablation)UniPixel-3B · ST-Evidence-Gen (100 samples)J&F: 25.6 -> 44.7Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy