AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Actor as Its Own Critic: Unifying Region Understanding and Localization via CycleGRPO

Xin Zhang, Haochen Wang, Yikang Zhou, Zhuochen Wang, Robby T. Tan, Xiangtai Li

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Instruction-tuned multimodal assistants · 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 cycle-consistency reward (CycleGRPO) on top of SAMTok instead of no RLSAMTok-4B · DLC-BenchAvg. score: 61.9 -> 67.7Improved
use cycle reward instead of standard GRPO with IoU reward on generated captions (ablation)SAMTok-4B · DLC-BenchAvg. score: 63.4 -> 67.7Improved
add CycleGRPO instead of standard GRPO (ablation)SAMTok-4B · GroundingSuitegIoU All: 62.7 -> 67.6Improved
add CycleGRPOSAMTok-4B · GRESN-acc (avg): 58.7 -> 92.1Improved
add standard GRPO (no cycle) to SAMTok (ablation)SAMTok-4B · GRESgIoU (avg): 71.2 -> 70.8Hurt
add CycleGRPOSAMTok-4B · GAR-Bench-VQAOverall accuracy: 64.2 -> 65.1Improved
add CycleGRPOSAMTok-4B · GAR-Bench-VQA (Position and Relation)accuracy: Position -3.2, Relation -1.0Hurt
add standard GRPO to SAMTok (ablation)SAMTok-4B · GAR-Bench-VQAOverall accuracy: 64.2 -> 63.9No clear effect
add CycleGRPOSAMTok-4B · GCG (Val)CIDEr: 48.2 -> 54.7Improved
use cycle reward instead of captioning-only GRPO with GT captions (ablation)SAMTok-4B · GroundingSuite (DAM-triples training)All: 56.9 -> 68.5Improved
use cycle reward instead of localization-only GRPO with GT captions (ablation)SAMTok-4B · DLC-Bench (DAM-triples training)Avg. score: 61.8 -> 68.0Improved
increase rollout counts G=K from 2 to 6 (ablation)CycleGRPO · GroundingSuiteAll: 62.5 -> 67.6Improved

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