GeoSearcher: Anchor-Guided Progressive Reasoning for Remote Sensing Visual Grounding with Process Supervision

2026-07-01

AI Summary

This work addresses the instability of existing one-step coordinate generation methods in remote sensing visual grounding, which stems from tiny targets, dense distractors, and multi-clue natural language queries. To overcome these challenges, the authors propose an anchor-guided progressive reasoning framework that reformulates localization as a two-stage process: first integrating localized cues via an anchor-centric reasoning mechanism, then refining predictions through a process-faithful strategy for structured reasoning. The approach innovatively incorporates process supervision and step-wise reasoning rewards, implemented through Anchor-Centric Reasoning Supervised Fine-Tuning (ACR-SFT) and Process-Faithful Group Relative Policy Optimization (PF-GRPO), which jointly leverage process-aware rewards and informative reasoning sample selection. Extensive experiments on DIOR-RSVG, OPT-RSVG, and VRS-Bench demonstrate significant performance gains over state-of-the-art methods, particularly enhancing localization accuracy for small objects and complex queries.

Abstract

Recent multimodal large language models (MLLMs) have shown strong cross-modal understanding and coordinate generation abilities in visual grounding. However, transferring these abilities to remote sensing visual grounding (RSVG) remains challenging. High-resolution remote sensing images usually cover large-scale scenes, where targets are often extremely small and surrounded by numerous visually similar distractors. Meanwhile, queries often contain multiple clues, such as reference objects, spatial relations, and target attributes. Existing MLLM-based methods usually formulate RSVG as one-step coordinate generation, which may lead to unstable predictions for small-object localization and complex queries. To address these challenges, we propose GeoSearcher, which reformulates RSVG as an anchor-guided progressive reasoning process and realizes it through two coupled stages: Anchor-Centric Reasoning Supervised Fine-Tuning (ACR-SFT) and Process-Faithful Group Relative Policy Optimization (PF-GRPO). In ACR-SFT, anchor-centric reasoning data are used to teach the model to represent key visual clues as anchors and progressively integrate location, relational, and attribute clues around them. In PF-GRPO, Process-Aware Reward (PAR) and Reasoning-Informative Sample Selector (RISS) further optimize this reasoning behavior by jointly evaluating key reasoning steps and target localization, while focusing training on samples that are more beneficial for improving progressive reasoning. Through this design, GeoSearcher transforms large-scale visual search into a more constrained local reasoning process. Extensive experiments on DIOR-RSVG, OPT-RSVG, and VRS-Bench show that GeoSearcher outperforms existing state-of-the-art methods. The project will be released at https://github.com/wangdianyu954-xixi/GeoSearcher.

Problem

• remote sensing visual grounding
• small object localization
• multimodal large language models
• complex query understanding
• visual distractors

Innovation

• anchor-guided reasoning
• progressive visual grounding
• process supervision
• remote sensing
• multimodal large language models