Aircrafts Dataset.

Training Patches. 
We collect 500 positive patches and 5000 negative patches from the Google Earth. The positive samples are image patches with an aircraft. The negative samples are randomly sampled patches of the airfield background. The original patch size is 64×64 pixels. 

Test dataset. 
There 26 test images. The image sizes vary from 565×369 to 1484×865 pixels. There are 453 aircrafts in the test dataset. The images were manually extracted from Google Earth.

Please cite the following paper when publishing results that use this dataset:
Hui Wu, Hui Zhang, Jinfang Zhang, Fanjiang Xu, "FAST AIRCRAFT DETECTION IN SATELLITE IMAGES BASED ON CONVOLUTIONAL NEURAL NETWORKS," ICIP,2015.

Hui Wu
Institute of Software Chinese Academy of Sciences, China
Email: wuhui13@iscas.ac.cn