AI Paper Atlas

Research map / Video / Video-language understanding

A.I.R.: Enabling Adaptive, Iterative, and Reasoning-based Frame Selection For Video Question Answering

Yuanhao Zou, Shengji Jin, Andong Deng, Youpeng Zhao, Jun Wang, Chen Chen

ICLR 2026 · Video › Video-language understanding · approach: Video question answering · 25 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
add A.I.R. frame selection over uniform samplingQwenVL-2.5-7B · Video-MME w/o subtitlesaccuracy (%): 60.8 -> 65.0Improved
add A.I.R. frame selection over uniform samplingQwenVL-2.5-7B · NextQAaccuracy (%): 74.3 -> 81.3Improved
use A.I.R. instead of MDP3 frame selectionInternVL-3-8B · MLVU devaccuracy (%): 74.0 -> 74.5Improved
add A.I.R. frame selection over uniform samplingInternVL-3-8B · NextQAaccuracy (%): 82.3 -> 82.6Improved
add A.I.R. frame selection over uniform samplingVILA-1.5-8B · Video-MME w/o subtitlesaccuracy (%): 48.9 -> 53.7Improved
add A.I.R. over MDP3QwenVL-2.5-7B · Egoschema Fullaccuracy (%): 56.8 -> 58.8Improved
raise max frame budget to 256 with A.I.R. versus uniform 256 (ablation)QwenVL-2.5-7B · Video-MME w/o subtitlesaccuracy (%): 63.4 -> 66.4Improved
use A.I.R. in place of BOLT at 32 framesLLaVA-OneVision-7B · Video-MME w/o subtitlesaccuracy (%): 59.9 -> 61.4Improved
use A.I.R. versus BOLT on Egoschema subsetLLaVA-OneVision-7B · Egoschema Subset (32 frames)accuracy (%): 64.0 -> 63.2Mixed
scale base VLM from 7B to larger QwenVL-2.5 models (ablation)A.I.R. · QwenVL-2.5 7B/32B/72Baccuracy gain: +4.2%Mixed
reduce worst-case VLM workload w_worst from 72 to 16 (ablation)A.I.R. · NextQAaccuracy (%): 82.6 -> 81.7Hurt
remove reasoning-based VLM analysis (ablation)A.I.R. · component ablationaccuracyImproved

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