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

• URL: http://arxiv.org/abs/2607.11581v1
• Date: Mon, 13 Jul 2026 14:00:57 GMT
• Title: Actor as Its Own Critic: Unifying Region Understanding and Localization via CycleGRPO
• Abstract summary: Cycle Group Relative Policy Optimization (CycleGRPO) jointly optimize region understanding and localization for Multimodal Large Language Models (MLLMs)<n>A single MLLM first acts as the actor to generate region captions, then immediately transitions to a critic to ground its generated text back in the spatial domain.<n>A quality-aware token-level cycle-consistency reward is employed to assess the semantic discriminability of text captions.
• Score: 43.173161414861966
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: This paper introduces Actor as Its Own Critic, a unified reinforcement learning framework, Cycle Group Relative Policy Optimization (CycleGRPO), that jointly optimizes region understanding and localization for Multimodal Large Language Models (MLLMs). Unlike existing separate pipelines, we leverage the inherent duality between the two tasks to construct a self-evaluating reinforcement learning paradigm: "region $\to$ text $\to$ region''. Specifically, a single MLLM first acts as the actor to generate region captions, then immediately transitions to a critic to ground its generated text back in the spatial domain. Therefore, CycleGRPO requires only region inputs, e.g., masks or bounding boxes, entirely bypassing the need for textual ground truths. A quality-aware token-level cycle-consistency reward is employed to assess the semantic discriminability of text captions via their physical localization accuracy. Empirically, built upon SAMTok, our CycleGRPO framework successfully bootstraps both capabilities simultaneously. Without any task-specific fine-tuning, the framework yields consistent performance gains across a wide range of benchmarks, including region captioning, region VQA, grounded dialogue, and referring segmentation. Overall, CycleGRPO offers a straightforward and scalable way to advance pixel-level capabilities in MLLMs. Code and models are released at https://github.com/devinxzhang/CycleGRPO.

Related papers

• From Coordinates to Candidate Regions: Temporal Change Localization via Region Selection in Remote Sensing Multimodal LLMs [48.93485452013392]
  multimodal large language models (RS-MLLMs) have advanced scene understanding and visual question answering over satellite imagery.<n>We present an RS-specific formulation of the region selection paradigm, previously explored in natural-image MLLMs.<n>Our framework employs a text-conditioned region proposal module, encodes each candidate as special tokens carrying per-frame visual features enriched with spatial and temporal cues.
  arXiv  Detail & Related papers  (2026-09-08T07:59:21Z)
• PixVL: Self-Supervised Training of Pixel-Level MLLMs via a Unified Mask--Text Consistency Cycle [31.234013315817858]
  PixVL is a self-supervised post-training framework for pixel-level multimodal large language models.<n>It generates and self-verify regional descriptions and learn from unlabeled data.<n> Experiments demonstrate that PixVL improves both region understanding and segmentation task.
  arXiv  Detail & Related papers  (2026-08-02T16:13:52Z)
• ROGLE: Robust Global-Local Alignment with Automated Region Supervision for Text-Based Person Search [12.859616063162491]
  Text-Based Person Search models struggle with fine-grained understanding due to global representational bias and semantic sparsity inherited from training on short captions.<n>We propose ROGLE, a unified framework that overcomes reliance on costly manual annotations through an automated Region-to-Sentence Matching strategy.<n>We also introduce the P-VLG Benchmark, a large-scale dataset constructed by curating and enriching images from established public benchmarks.
  arXiv  Detail & Related papers  (2026-06-01T07:41:44Z)
• Seg-Agent: Test-Time Multimodal Reasoning for Training-Free Language-Guided Segmentation [52.8308168727975]
  Seg-Agent is a completely training-free framework that pioneers Explicit Multimodal Chain-of-Reasoning.<n>Our approach constructs an interactive visual reasoning loop comprising three stages: generation, selection, and refinement.<n>Various-LangSeg is a novel benchmark covering explicit semantic, generic object, and reasoning-guided segmentation tasks.
  arXiv  Detail & Related papers  (2026-05-13T03:36:44Z)
• Qwen3-VL-Seg: Unlocking Open-World Referring Segmentation with Vision-Language Grounding [26.30521907946121]
  We present Qwen3-VL-Seg, a parameter-efficient framework that treats the MLLM-predicted box as a semantically grounded structural prior.<n>At its core, a lightweight box-guided mask decoder combines multi-scale spatial feature injection, spatial-semantic query construction, box-guided high-resolution pixel fusion.<n>Experiments on referring expression segmentation, visual grounding, and ORS-Bench show that Qwen3-VL-Seg performs strongly across closed-set and open-world settings.
  arXiv  Detail & Related papers  (2026-05-08T02:20:40Z)
• Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning [63.109585527799005]
  GroundingAgent is a visual grounding framework that operates without task-specific fine-tuning.<n>It achieves an average zero-shot grounding accuracy of 65.1 % on widely-used benchmarks.<n>It also offers strong interpretability, transparently illustrating each reasoning step.
  arXiv  Detail & Related papers  (2025-11-24T03:11:08Z)
• FineRS: Fine-grained Reasoning and Segmentation of Small Objects with Reinforcement Learning [62.11389260206383]
  textscFineRS is a two-stage MLLM-based reinforcement learning framework for segmenting extremely small objects.<n>We present textscFineRS-4k, a new dataset for evaluating MLLMs on attribute-level reasoning and pixel-level segmentation on subtle, small-scale targets.
  arXiv  Detail & Related papers  (2025-10-24T10:14:17Z)
• ESCA: Contextualizing Embodied Agents via Scene-Graph Generation [47.008144510161486]
  We propose ESCA, a framework that contextualizes embodied agents by grounding their perception in spatial-temporal scene graphs.<n>At its core is SGCLIP, a novel, open-domain, promptable foundation model for generating scene graphs.<n>SGCLIP excels in both prompt-based inference and task-specific fine-tuning, achieving state-of-the-art results on scene graph generation and action localization benchmarks.
  arXiv  Detail & Related papers  (2025-10-11T20:13:59Z)
• URECA: Unique Region Caption Anything [29.363967361960043]
  Region-level captioning aims to generate natural language descriptions for specific image regions while highlighting their distinguishing features.<n>We introduce URECA dataset, a large-scale dataset tailored for multi-granularity region captioning.<n>We present URECA, a novel captioning model designed to effectively encode multi-granularity regions.
  arXiv  Detail & Related papers  (2025-04-07T17:59:44Z)
• Global and Local Semantic Completion Learning for Vision-Language Pre-training [34.740507502215536]
  Cross-modal alignment plays a crucial role in vision-language pre-training models. We propose a novel Global and Local Semantic Completion Learning (GLSCL) task to facilitate global-local alignment and local-local alignment simultaneously.
  arXiv  Detail & Related papers  (2023-06-12T13:20:29Z)
• Context-aware Biaffine Localizing Network for Temporal Sentence Grounding [61.18824806906945]
  This paper addresses the problem of temporal sentence grounding (TSG) TSG aims to identify the temporal boundary of a specific segment from an untrimmed video by a sentence query. We propose a novel localization framework that scores all pairs of start and end indices within the video simultaneously with a biaffine mechanism.
  arXiv  Detail & Related papers  (2021-03-22T03:13:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.