MedVol-R1: Reward-Driven Evidence Grounding for Volumetric Reasoning Segmentation 文章

ArXiv CS.CV2026-05-27NEWSen作者: Zichun Wang, Hairong Shi, Bingzheng Wei, Yan Xu, Zihua Wang
查看原文 →
关系图谱

MedVol-R1: Reward-Driven Evidence Grounding for Volumetric Reasoning Segmentation · 相关技术

相关技术

ORM远程代码执行（RCE）reinforcement learninglatent representationinterpretabilityZenVerseTokenRepresentation AnchoringQueryOWLMITLayout-as-PolicyLarge vision-language modelVLMHISGrouped Memorization EvaluationForDual-Stream Attention ModuleDSACodingCamouflaged object detectionCLICISARGANNICE