AI Paper Atlas

Research map / Video / Video-language understanding

Are Video Reasoning Models Ready to Go Outside?

Yangfan He, Changgyu Boo, Jaehong Yoon

ECCV 2026 · Video › Video-language understanding · approach: Event and anomaly reasoning in video · 2 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 ROVA training to Qwen2.5-VL-7BQwen2.5-VL-7B · PVRBench (perturbed, avg)accuracy: .33 -> .47Improved
add ROVA training to Embodied-REmbodied-R 7B · PVRBench (perturbed, avg)accuracy: 0.42 -> 0.50Improved
add ROVA training to Qwen2.5-VL-72BQwen2.5-VL-72B · PVRBench (perturbed, avg)accuracy: .45 -> .56Improved
add ROVA training (effect on clean data)Qwen2.5-VL-7B · PVRBench (clean, Orig.)accuracy: .51 -> .53Improved
add curriculum (SRE+DRE+ME) to dual-branch alignment (ablation)ROVA · PVRBench (avg acc, training efficiency)GPU-hours and accuracy: 142.8 -> 134.4 GPU-h; .48 -> .53 (text: 0.37 to 0.47)Improved
add dual-branch structured corruption alignment over standard GRPO (ablation)Qwen2.5-VL-7B · PVRBench (avg acc, training efficiency)accuracy: .45 -> .48Improved
difficulty-aware discarding instead of random discarding (ablation)ROVA · PVRBenchaccuracy: up to +3.4%Improved
replace GPT-4o reward judge with open-source Qwen3-13B (ablation)ROVA · PVRBenchaccuracy: 0.470 -> 0.467No clear effect
use step-level or conditional alignment reward (ablation)ROVA · PVRBenchaccuracyHurt
structured perturbation-aware masks instead of random masking (ablation)ROVA · PVRBenchaccuracy: 6-9% absoluteImproved
increase perturbation intensity from 0.7 to 0.9 (ablation)ROVA · PVRBench valavg accuracy (%): 39.1 -> 36.8Hurt
lower perturbation intensity to 0.5 (ablation)ROVA · PVRBench valavg accuracy (%): 39.1 -> 40.2Mixed

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