Cars in Stuttgart

This dataset is an educational toy dataset for object detection with remote sensing data. The dataset contains aerial imagery of the city of Stuttgart at 20cm ground sample distance together with bounding box labels indicating cars (single class) present in these images. Labels are present in the COCO format.

The dataset is split into three parts:* a training split (1000 images),* a validation split (190 images) and* a test split (190 images).

Each image features a height and width of 128 pixels each, contains RGB bands and is stored in the png format. Image data were cropped from aerial imagery of the city of Stuttgart taken in 2021, which are available through Stadtmessungsamt Stuttgart as open data under the CC BY 4.0 license: https://opendata.stuttgart.de/dataset/luftbilder-2021

This dataset has been labeled by Yilsey Terea Benavides Miranda, Khem Raj Devkota, David Michael Udoh and Gökhan Yücesan as part of the Master of Photogrammetry and Geoinformatics course "Remote Sensing Studio" in the 2024 summer term at the Stuttgart University of Applied Sciences.