A2Seek: Towards Reasoning-Centric Benchmark for Aerial Anomaly Understanding

TL;DR

This paper introduces A2Seek, a comprehensive aerial anomaly dataset with detailed annotations and a reasoning framework, A2Seek-R1, that significantly improves anomaly detection and localization in drone-view scenarios.

Contribution

The paper presents A2Seek, a large-scale aerial anomaly dataset with rich annotations, and proposes A2Seek-R1, a reasoning-based framework that enhances understanding of anomalies in aerial videos.

Findings

A2Seek-R1 improves AP by up to 22.04% for prediction accuracy.

A2Seek-R1 achieves a 13.9% gain in mIoU for anomaly localization.

The framework generalizes well across complex and out-of-distribution environments.

Abstract

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/.