Asking like Socrates: Socrates helps VLMs understand remote sensing images

TL;DR

This paper introduces RS-EoT, an iterative, evidence-seeking reasoning paradigm for remote sensing vision-language tasks, addressing pseudo reasoning caused by the Glance Effect and achieving state-of-the-art results.

Contribution

It proposes a novel SocraticAgent system with a two-stage reinforcement learning strategy to improve genuine evidence-based reasoning in remote sensing models.

Findings

RS-EoT achieves state-of-the-art performance on multiple benchmarks.

The approach mitigates the Glance Effect, enabling more accurate reasoning.

Iterative reasoning cycles are confirmed through analysis.

Abstract

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