Awesome

HRPlanes Dataset Details

The imagery required for the dataset has been obtained from Google Earth. We have downloaded 4800 x 2703 sized 3092 RGB images from the biggest airports of the world such as Paris-Charles de Gaulle, John F. Kennedy, Frankfurt, Istanbul, Madrid, Dallas, Las Vegas and Amsterdam Airports and aircraft boneyards like Davis-Monthan Air Force Base. Dataset images were annotated manually by creating bounding boxes for each airplane using formerly HyperLabel. Quality control of each label was conducted by visual inspection of independent analysts who were not included in the labelling procedure. A total of 18,477 airplanes have been labelled. A sample image and corresponding minimum boxes for airplanes can be seen the figure. The dataset has been approximately split as 70% (2166 images), 20% (615 images) and 10% (311 images) for training, validation and testing, respectively.

Download The Dataset

You can now download the entire dataset in YOLO format from Zenodo.

If you make use of the HRPlanes dataset, please cite our paper.

Download The Weights

The weights for YOLOv4 and Faster R-CNN are available here.

Citation

If you make use of the test dataset or weights, please cite our paper.

Bakırman, T. & Sertel, E. (2023). "A benchmark dataset for deep learning-based airplane detection: HRPlanes". International Journal of Engineering and Geosciences , 8 (3) , 212-223 . DOI: 10.26833/ijeg.1107890.

Copyright

Use of the Google Earth images must respect the "Google Earth" terms of use. All images and their associated annotations can be used for academic purposes only, but any commercial use is prohibited.

Released under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) License.