AI Paper Atlas

Research map / Video / Video-language understanding

APPO: Attention-guided Perception Policy Optimization for Video Reasoning

Henghui Du, Chang Zhou, Xi Chen, Di Hu

CVPR 2026 · Video › Video-language understanding · approach: Event and anomaly reasoning in video · 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
replace GRPO/DAPO with APPO (attention-guided token re-weighting)Qwen2.5-VL-3B · SEED-Bench-R1 L3 (OOD)accuracy: 31.8 -> 35.0 (vs DAPO)Improved
replace DAPO with APPOQwen2.5-VL-7B · SEED-Bench-R1 L1 (in-dist)accuracy: 50.0 -> 50.5Improved
replace DAPO with APPOQwen2.5-VL-3B · NExT-GQAmIoU: 10.6 -> 11.1Improved
replace DAPO with APPOQwen2.5-VL-3B · MVBenchaccuracy: 62.8 -> 62.4No clear effect
replace DAPO with APPOQwen2.5-VL-7B · NExT-QAaccuracy: 79.9 -> 79.6No clear effect
RL (GRPO) instead of SFTQwen2.5-VL-7B · SEED-Bench-R1 L1 (in-dist)accuracy: 40.2 -> 49.0Improved
enhance perception model from Qwen2.5-VL-7B to 32B (ablation)Qwen3-8B reasoner · SEED-Bench-R1 L1accuracy: +1.4%Improved
enhance reasoning model from Qwen3-8B to OpenAI-o3 (ablation)Qwen2.5-VL-7B perception · SEED-Bench-R1 L1accuracy: +0.7%No clear effect
increase K1 (tokens per response for attention aggregation) (ablation)APPO · SEED-Bench-R1 (Qwen2.5-VL-7B)accuracyImproved
increase K2 beyond 3 (selected frames per response) (ablation)APPO · SEED-Bench-R1 (Qwen2.5-VL-7B)accuracyHurt
increase K3 (perception tokens per response) to large values (ablation)APPO · SEED-Bench-R1 (Qwen2.5-VL-7B)accuracyHurt
vary token weight scale alpha from 1.2 to 2.0 (ablation)APPO · SEED-Bench-R1 (Qwen2.5-VL-7B)accuracyMixed

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