A2Seek: Towards Reasoning-Centric Benchmark for Aerial Anomaly Understanding

Haosheng Chen (Chongqing University of Post and Telecommunications) · Mengjingcheng Mo (Chongqing University of Posts and Telecommunications) · Xinyang Tong (Chong Qing University of Posts and Techcommunications) · Mingpi Tan (Chongqing University of Post and Telecommunications) · Jiaxu Leng (Chongqing University of Post and Telecommunications) · JianKang Zheng (Chongqing University of Post and Telecommunications) · Yiran Liu (Chongqing University of Post and Telecommunications) · Ji Gan (Chongqing University of Post and Telecommunications) · Weisheng Li (Chongqing Post and Communications University) · Xinbo Gao (Chongqing University of Post and Telecommunications)

aerial anomaly understandingaerial group relative policy optimizationanomaly detection anomaly localization causal reasoning generalizationgraph-of-thoughthigh-resolution aerial videosout-of-distribution scenarios prediction accuracyreasoning-centric benchmarkrule-based reward functionsseeking mechanismsupervised fine-tuninguavs

While unmanned aerial vehicles (UAVs) offer wide-area, high-altitude coverage for anomaly detection, they face challenges such as dynamic viewpoints, scale variations, and complex scenes. Existing datasets and methods, mainly designed for fixed ground-level views, struggle to adapt to these conditions, leading to significant performance drops in drone-view scenarios.To bridge this gap, we introduce A2Seek (Aerial Anomaly Seek), a large-scale, reasoning-centric benchmark dataset for aerial anomaly understanding. This dataset covers various scenarios and environmental conditions, providing high-resolution real-world aerial videos with detailed annotations, including anomaly categories, frame-level timestamps, region-level bounding boxes, and natural language explanations for causal reasoning. Building on this dataset, we propose A2Seek-R1, a novel reasoning framework that generalizes R1-style strategies to aerial anomaly understanding, enabling a deeper understanding of “Where” anomalies occur and “Why” they happen in aerial frames.To this end, A2Seek-R1 first employs a graph-of-thought (GoT)-guided supervised fine-tuning approach to activate the model's latent reasoning capabilities on A2Seek. Then, we introduce Aerial Group Relative Policy Optimization (A-GRPO) to design rule-based reward functions tailored to aerial scenarios. Furthermore, we propose a novel “seeking” mechanism that simulates UAV flight behavior by directing the model's attention to informative regions.Extensive experiments demonstrate that A2Seek-R1 achieves up to a 22.04\% improvement in AP for prediction accuracy and a 13.9\% gain in mIoU for anomaly localization, exhibiting strong generalization across complex environments and out-of-distribution scenarios. Our dataset and code are released at https://2-mo.github.io/A2Seek/.