DFC15 multilabel dataset is produced from GRSS_DFC_2015. Details are introduced in this paper. In total, there are 3342 aerial images (2673 training and 669 test) and 8 object labels. The size of each image is 600 x 600 pixels. The proposed dataset is now available on Google Drive.
If you find this dataset is useful, please kindly cite the following:
@article{hua2019recurrently,
title={Recurrently exploring class-wise attention in a hybrid convolutional and bidirectional {LSTM} network for multi-label aerial image classification},
author={Hua, Yuansheng and Mou, Lichao and Zhu, Xiao Xiang},
journal={ISPRS Journal of Photogrammetry and Remote Sensing},
volume={149},
pages={188--199},
year={2019},
}