AOD 4: Dataset for Air Borne Object Detection

The detection of airborne objects has largely been unexplored or is only limited to drones. This dataset provides a huge set of airborne objects, comprising 22,516 images categorised into four classes namely; airplanes, helicopters, drones, and birds. The dataset is made so as to include complex backgrounds to showcase the varying environmental conditions present in the real world. AOD-4 dataset consists of images with sky, cloudy, terrestrial backgrounds, as well as nighttime, rainy, and snowy conditions. The dataset also includes instances where multiple objects are present in the frame aiming towards Multi-Object Detection capabilities. Researchers can leverage this dataset to develop and refine algorithms for airborne object detection and tracking, with potential applications spanning military surveillance, border security, and public safety.

空中目标检测领域在很大程度上尚未得到充分探索，相关研究目前仅局限于无人机场景。本数据集收录了海量空中目标样本，共计22516张图像，共分为四大类别，分别为固定翼飞机（airplanes）、直升机（helicopters）、无人机（drones）与鸟类（birds）。本数据集特意设置复杂背景，以还原真实世界中多样化的环境条件。AOD-4数据集涵盖了带有天空、多云、陆地背景的图像，同时包含夜间、降雨、降雪等多种时段与气象场景。此外，数据集中还包含多目标同框的样本，以适配多目标检测（Multi-Object Detection）的算法研发需求。研究人员可依托本数据集开发并优化空中目标检测与跟踪算法，其潜在应用场景覆盖军事侦察、边境安防与公共安全领域。