CORE-Seg: Reasoning-Driven Segmentation for Complex Lesions via Reinforcement Learning

• URL: http://arxiv.org/abs/2603.05911v1
• Date: Fri, 06 Mar 2026 04:58:55 GMT
• Title: CORE-Seg: Reasoning-Driven Segmentation for Complex Lesions via Reinforcement Learning
• Abstract summary: We introduce ComLesion-14K, the first diverse Chain-of-Thought benchmark for reasoning-driven complex lesion segmentation.<n>We propose CORE-Seg, an end-to-end framework integrating reasoning with segmentation through a Semantic-Guided Prompt Adapter.<n>Our Method achieves state-of-the-art results with a mean Dice of 37.06% (14.89% higher than the second-best baseline), while reducing the failure rate to 18.42%.
• Score: 48.77689764063103
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Medical image segmentation is undergoing a paradigm shift from conventional visual pattern matching to cognitive reasoning analysis. Although Multimodal Large Language Models (MLLMs) have shown promise in integrating linguistic and visual knowledge, significant gaps remain: existing general MLLMs possess broad common sense but lack the specialized visual reasoning required for complex lesions, whereas traditional segmentation models excel at pixel-level segmentation but lack logical interpretability. In this paper, we introduce ComLesion-14K, the first diverse Chain-of-Thought (CoT) benchmark for reasoning-driven complex lesion segmentation. To accomplish this task, we propose CORE-Seg, an end-to-end framework integrating reasoning with segmentation through a Semantic-Guided Prompt Adapter. We design a progressive training strategy from SFT to GRPO, equipped with an adaptive dual-granularity reward mechanism to mitigate reward sparsity. Our Method achieves state-of-the-art results with a mean Dice of 37.06\% (14.89\% higher than the second-best baseline), while reducing the failure rate to 18.42\%. Project Page: https://xyxl024.github.io/CORE-Seg.github.io/

Related papers

• DGSeg: Dynamic Gating of Semantic-Spatial Guided Predictions for Reasoning Segmentation [5.5825182710791275]
  Reasoning segmentation aims to predict pixel-wise masks for targets given complex language queries.<n>We propose DGSeg, a framework that learns to fuse predictions guided by semantic and spatial cues.<n>Experiments demonstrate that DGSeg consistently outperforms strong baselines on multiple benchmarks.
  arXiv  Detail & Related papers  (2026-07-06T08:10:03Z)
• CR-Seg: Attention-Guided and CoT-Enhanced Coarse-to-Refined Reasoning Segmentation [26.21876180083083]
  Reasoning segmentation aims to segment target objects described by complex language through joint visual-textual reasoning.<n>Existing methods rely on either learned semantic tokens to bridge Multimodal Large Language Models (MLLMs) and segmentation models, suffering from difficult cross-modal alignment.<n>We propose Attention-Guided and CoT-Enhanced Coarse-to-Refined Reasoning, termed CR-Seg, a two-stage framework for coarse-to-refined spatial segmentation.
  arXiv  Detail & Related papers  (2026-06-02T12:30:04Z)
• Milestones over Outcome: Unlocking Geometric Reasoning with Sub-Goal Verifiable Reward [67.00373428443879]
  We introduce a paradigm shift towards subgoal-level evaluation and learning.<n>We first construct GeoGoal, a benchmark synthesized via a rigorous formal verification data engine.<n>We propose the Sub-Goal Verifiable Reward (SGVR) framework, which replaces sparse signals with dense rewards based on the Skeleton Rate.
  arXiv  Detail & Related papers  (2026-01-08T16:17:56Z)
• Bridging Semantics and Geometry: A Decoupled LVLM-SAM Framework for Reasoning Segmentation in Remote Sensing [8.731693840957716]
  Think2Seg-RS is a framework that trains an LVLM prompter to control a frozen Segment Anything Model (SAM) via structured geometric prompts.<n>The framework achieves state-of-the-art performance on the EarthReason dataset.<n> compact segmenters outperform larger ones under semantic-level supervision, and that negative prompts are ineffective in heterogeneous aerial backgrounds.
  arXiv  Detail & Related papers  (2025-12-22T11:46:42Z)
• Through the Looking Glass: A Dual Perspective on Weakly-Supervised Few-Shot Segmentation [46.635612270422655]
  Meta-learning aims to uniformly sample homogeneous support-query pairs, characterized by the same categories and similar attributes.<n>This identical network design results in over-semantic homogenization.<n>We propose a novel but heterogeneous network to enhance complementarity and preserve semantic commonality.<n>In the weakly-supervised few-shot semantic segmentation (WFSS) task, TLG achieves a 13.2% improvement on Pascal-5textsuperscripti and a 9.7% improvement on COCO-20textsuperscripti.
  arXiv  Detail & Related papers  (2025-08-22T07:29:30Z)
• LENS: Learning to Segment Anything with Unified Reinforced Reasoning [38.582392908238866]
  We introduce LENS, a scalable reinforcement-learning framework that jointly optimize the reasoning process and segmentation in an end-to-end manner.<n>LENS achieves an average cIoU of 81.2% on the RefCOCO, RefCOCO+, and RefCOCOg benchmarks, outperforming the strong fine-tuned method, i.e., GLaMM, by up to 5.6%.
  arXiv  Detail & Related papers  (2025-08-19T17:59:53Z)
• LIRA: Inferring Segmentation in Large Multi-modal Models with Local Interleaved Region Assistance [54.683384204063934]
  Large multi-modal models (LMMs) struggle with inaccurate segmentation and hallucinated comprehension.<n>We propose LIRA, a framework that capitalizes on the complementary relationship between visual comprehension and segmentation.<n>LIRA achieves state-of-the-art performance in both segmentation and comprehension tasks.
  arXiv  Detail & Related papers  (2025-07-08T07:46:26Z)
• DeRIS: Decoupling Perception and Cognition for Enhanced Referring Image Segmentation through Loopback Synergy [20.581841892290672]
  We propose DeRIS, a novel framework that decomposes RIS into two key components: perception and cognition.<n>Our findings reveal that the predominant limitation lies not in perceptual deficiencies, but in the insufficient multi-modal cognitive capacity of current models.<n>We introduce a simple non-referent sample conversion data augmentation to address the long-tail distribution issue related to target existence judgement.
  arXiv  Detail & Related papers  (2025-07-02T14:14:35Z)
• Think Before You Segment: High-Quality Reasoning Segmentation with GPT Chain of Thoughts [64.93416171745693]
  ThinkFirst is a training-free reasoning segmentation framework.<n>Our approach allows GPT-4o or other powerful MLLMs to generate a detailed, chain-of-thought description of an image.<n>This summarized description is then passed to a language-instructed segmentation assistant to aid the segmentation process.
  arXiv  Detail & Related papers  (2025-03-10T16:26:11Z)
• Learning What Not to Segment: A New Perspective on Few-Shot Segmentation [63.910211095033596]
  Recently few-shot segmentation (FSS) has been extensively developed. This paper proposes a fresh and straightforward insight to alleviate the problem. In light of the unique nature of the proposed approach, we also extend it to a more realistic but challenging setting.
  arXiv  Detail & Related papers  (2022-03-15T03:08:27Z)
• DONet: Dual Objective Networks for Skin Lesion Segmentation [77.9806410198298]
  We propose a simple yet effective framework, named Dual Objective Networks (DONet), to improve the skin lesion segmentation. Our DONet adopts two symmetric decoders to produce different predictions for approaching different objectives. To address the challenge of large variety of lesion scales and shapes in dermoscopic images, we additionally propose a recurrent context encoding module (RCEM)
  arXiv  Detail & Related papers  (2020-08-19T06:02:46Z)

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.