用强化学习让大模型轻松搞定像素级分割，零样本泛化能力强。
Seg-R1: Segmentation Can Be Surprisingly Simple with Reinforcement Learning
- 用强化学习生成点框提示，引导SAM2完成分割
- 纯强化学习训练下COD10K S-measure达0.873
- 无需文本监督，零样本表现超越全监督模型
我们提出Seg-R1，首次探索利用强化学习（RL）增强大模型在像素级理解与推理方面的能力。以前景分割任务为例，包括伪装物体检测（COD）和显著物体检测（SOD），该方法让大模型以逐令牌方式生成点和边界框提示，进而指导SAM2生成分割掩码。我们引入组相对策略优化（GRPO）至分割领域，通过精心设计的训练策略赋予大模型像素级理解能力。值得注意的是，仅使用强化学习训练的Seg-R1在无需复杂模型修改的情况下，于COD10K数据集上达到0.873的S-measure。此外，纯强化学习训练展现出强大开放世界泛化能力：尽管仅在前景分割图像-掩码对上训练且无文本监督，其在指代分割（RefCOCOg测试集）和推理分割（ReasonSeg测试集）上分别取得71.4 cIoU和56.7 gIoU，均优于在这些数据集上全监督训练的模型。
原文摘要 · Abstract (English)
We present Seg-R1, a preliminary exploration of using reinforcement learning (RL) to enhance the pixel-level understanding and reasoning capabilities of large multimodal models (LMMs). Starting with foreground segmentation tasks, specifically camouflaged object detection (COD) and salient object detection (SOD), our approach enables the LMM to generate point and bounding box prompts in the next-token fashion, which are then used to guide SAM2 in producing segmentation masks. We introduce Group Relative Policy Optimization (GRPO) into the segmentation domain, equipping the LMM with pixel-level comprehension through a carefully designed training strategy. Notably, Seg-R1 achieves remarkable performance with purely RL-based training, achieving .873 S-measure on COD10K without complex model modification. Moreover, we found that pure RL training demonstrates strong open-world generalization. Despite being trained solely on foreground segmentation image-mask pairs without text supervision, Seg-R1 achieves impressive zero-shot performance on referring segmentation and reasoning segmentation tasks, with 71.4 cIoU on RefCOCOg test and 56.7 gIoU on ReasonSeg test, outperforming models fully supervised on these datasets.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。