像苏格拉底提问：苏格拉底帮助VLMs理解遥感图像
Asking like Socrates: Socrates helps VLMs understand remote sensing images
AI总结：
本文提出RS-EoT方法，通过迭代视觉证据寻求机制和两阶段强化学习策略，解决遥感图像中伪推理问题，提升视觉证据基础的推理能力。
摘要：
近期多模态推理模型受DeepSeek-R1启发，显著提升了视觉语言系统。然而在遥感（RS）任务中，我们观察到广泛存在的伪推理现象：模型叙述推理过程而非基于视觉证据真实推理。我们将其归因于“Glance Effect”，即对大规模遥感图像的单次粗略感知导致基于语言自洽而非视觉证据的不完整理解与推理。为此，我们提出RS-EoT（遥感证据-思维）语言驱动、迭代视觉证据寻求范式。为植入此范式，我们提出SocraticAgent，一种自博弈多智能体系统，通过交替的推理与视觉检查循环合成推理轨迹。为增强和推广这些模式，我们提出两阶段渐进式强化学习策略：首先在细粒度Grounding任务上进行强化学习以增强RS-EoT能力，随后在RS VQA上进行强化学习以推广至更广泛的理解场景。实验表明RS-EoT在多个RS VQA和Grounding基准测试中取得最优性能。分析显示清晰的推理与证据寻求迭代循环，证实RS-EoT缓解了Glance Effect并实现了真实证据基础的推理。我们的代码、数据和模型可在https://geox-lab.github.io/Asking_like_Socrates获取。
英文摘要：
Recent multimodal reasoning models, inspired by DeepSeek-R1, have significantly advanced vision-language systems. However, in remote sensing (RS) tasks, we observe widespread pseudo reasoning: models narrate the process of reasoning rather than genuinely reason toward the correct answer based on visual evidence. We attribute this to the Glance Effect, where a single, coarse perception of large-scale RS imagery results in incomplete understanding and reasoning based on linguistic self-consistency instead of visual evidence. To address this, we propose RS-EoT (Remote Sensing Evidence-of-Thought), a language-driven, iterative visual evidence-seeking paradigm. To instill this paradigm, we propose SocraticAgent, a self-play multi-agent system that synthesizes reasoning traces via alternating cycles of reasoning and visual inspection. To enhance and generalize these patterns, we propose a two-stage progressive RL strategy: first, RL on fine-grained Grounding tasks to enhance RS-EoT capabilities, followed by RL on RS VQA to generalize to broader understanding scenarios. Experiments show RS-EoT achieves state-of-the-art performance on multiple RS VQA and grounding benchmarks. Analyses reveal clear iterative cycles of reasoning and evidence seeking, confirming RS-EoT mitigates the Glance Effect and enables genuine evidence-grounded reasoning. Our code, data, and models are available at https://geox-lab.github.io/Asking_like_Socrates
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
