Crop-Tool-Augmented Active Perception with Reinforcement Learning for High-Resolution Remote Sensing Visual Question Answering
Remote Sensing · 08.07.2026 · 18 (14) · 2288
DOI: 10.3390/rs18142288 ↗Kurz gefasst
Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 08.07.2026. Verfasst von Qian Li und 3 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Beijing Institute of Technology.
Abstract (Originalsprache)
High-resolution remote sensing visual question answering (RS-VQA) requires models to identify question-relevant regions and reason over fine-grained visual evidence. However, existing vision–language models usually rely on fixed global image inputs, which may lose critical local details in ultra-high-resolution imagery and struggle with sparse informative regions, large object-scale variations, and complex spatial layouts. To address these challenges, this paper proposes a crop-tool-augmented active perception framework with reinforcement learning. The framework introduces structured tokens to explicitly organize the reasoning process into question understanding, cropping decision-making, local evidence acquisition and final answer generation. Based on this design, the model can actively determine whether a cropping operation is needed and select task-relevant regions for further inspection. To enable stable tool-use and multi-turn reasoning in a compact vision–language model, we construct teacher-guided cropping reasoning trajectories from high-resolution images, question–answer pairs, and annotated regions in the LRS-GRO dataset, and use them for cold-start supervised fine-tuning of Qwen2.5-VL-3B. Furthermore, we introduce Group Relative Policy Optimization to refine the model’s active perception policy. A region-aware reward function is designed by integrating output-format constraints, reference-region coverage, answer semantic consistency, and cropping penalties, which encourages compact and informative region selection while reducing redundant tool invocations. Experiments on VRSBench, MME-RealWorld-RS, XLRS-Bench, and LRS-VQA demonstrate that the proposed method achieves competitive overall performance compared with closed-source, open-source, and remote-sensing-specific vision–language models, and obtains the best or comparable results on most benchmarks. Ablation studies further verify the effectiveness of structured supervised fine-tuning, reinforcement learning optimization, and the proposed reward design.
Förderung
- Beijing Institute of Technology
Zitieren
APA
Li, Q., Chen, K., Han, Y., & Xu, X. (2026). Crop-Tool-Augmented Active Perception with Reinforcement Learning for High-Resolution Remote Sensing Visual Question Answering. Remote Sensing, 18(14), 2288. https://doi.org/10.3390/rs18142288
Vancouver
Li Q, Chen K, Han Y, Xu X. Crop-Tool-Augmented Active Perception with Reinforcement Learning for High-Resolution Remote Sensing Visual Question Answering. Remote Sensing. 2026;18(14):2288. doi:10.3390/rs18142288
BibTeX
@article{Li2026,
author = {Li, Qian and Chen, Kailiang and Han, Yitong and Xu, Xiangyang},
title = {Crop-Tool-Augmented Active Perception with Reinforcement Learning for High-Resolution Remote Sensing Visual Question Answering},
journal = {Remote Sensing},
year = {2026},
volume = {18},
number = {14},
pages = {2288},
doi = {10.3390/rs18142288}
}
EUVIMED