Aerial Person Detection Test Set

Aerial Person Test Dataset is a held-out dataset designed for the evaluation of person detection models in aerial and top-view imagery. The dataset contains 500 images and their corresponding object detection annotations in YOLO format. It is intended exclusively as a test set. Classes The dataset contains a single class: 0 - person Each image has a corresponding .txt annotation file with the same filename. Every annotation line follows the YOLO format: class_id x_center y_center width height The bounding-box coordinates are normalized to the interval [0, 1].Data collection and annotation The images were collected from aerial and top-view visual sources, including Pexels, Vecteezy, YouTube live camera streams, and SkylineWebcams. The images were initially pre-annotated using a YOLO-based person detection pipeline. The generated bounding boxes were subsequently manually reviewed, corrected, and completed using Label Studio, before being exported in YOLO format. Dataset structure aerial_person_test_dataset/├── images/│ └── test/├── labels/│ └── test/├── data.yaml├── README.md├── SOURCES.md└── RIGHTS.md Dataset split Test: 500 images Usage The dataset can be used with YOLO-based object detection frameworks through the included data.yaml configuration file. Example command: yolo detect val data=data.yaml model=yolo11m.pt imgsz=1280 split=test Access and reuse Access to the dataset files is restricted and access requests are evaluated individ ually by the record owners. The use of the dataset must comply with the licenses, copyright conditions, and terms of use of the original data sources. Redistribution of the dataset or any of its components is not permitted without prior approval from the record owners. Further information regarding the original sources and applicable usage conditions is provided in the included SOURCES.md and RIGHTS.md files.