用强化学习实现视频对象分割的可解释推理，提升复杂场景泛化能力。
VideoSeg-R1:Reasoning Video Object Segmentation via Reinforcement Learning
- 分阶段设计：采样、推理、分割传播，模拟人类注意力与逻辑链
- 在多个基准上达到顶尖性能，任务难度自适应控制推理长度
- 适合需要可解释性与强泛化的视频理解研究者
传统视频推理分割方法依赖监督微调，限制了对分布外场景的泛化能力且缺乏显式推理过程。为此，我们提出首个将强化学习引入视频推理分割的框架 VideoSeg-R1。该框架采用解耦架构，将任务分解为联合指代图像分割与视频掩码传播。包含三个阶段：(1) 分层文本引导帧采样器，模拟人类注意力；(2) 推理模型生成空间提示并输出显式推理链；(3) 利用 SAM2 与 XMem 完成分割与传播。引入任务难度感知机制，动态调节推理长度以提升效率与精度。在多个基准上的大量实验表明，VideoSeg-R1 在复杂视频推理与分割任务中达到当前最优表现。代码将公开于 https://github.com/euyis1019/VideoSeg-R1。
原文摘要 · Abstract (English)
Traditional video reasoning segmentation methods rely on supervised fine-tuning, which limits generalization to out-of-distribution scenarios and lacks explicit reasoning. To address this, we propose \textbf{VideoSeg-R1}, the first framework to introduce reinforcement learning into video reasoning segmentation. It adopts a decoupled architecture that formulates the task as joint referring image segmentation and video mask propagation. It comprises three stages: (1) A hierarchical text-guided frame sampler to emulate human attention; (2) A reasoning model that produces spatial cues along with explicit reasoning chains; and (3) A segmentation-propagation stage using SAM2 and XMem. A task difficulty-aware mechanism adaptively controls reasoning length for better efficiency and accuracy. Extensive evaluations on multiple benchmarks demonstrate that VideoSeg-R1 achieves state-of-the-art performance in complex video reasoning and segmentation tasks. The code will be publicly available at https://github.com/euyis1019/VideoSeg-R1.
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。