In this repository, we try to use deep learning method to detect power plants in remote sensing images. We provide a dataset which contains 892 remote sensing images. We also offer eight deep learning models which contain Faster R-CNN, R-FCN, FPN, DCN, SSD, DSSD, YOLO and RetinaNet.
Our dataset which is collected from Google Earth covers 60 fossil fuel power plants in the Beijing-Tianjin-Hebei Region. We divide the dataset into four classes i.e. working chimney, unworking chimney, working condensing tower and unworking condensing tower. The format of annotations and label are pascal VOC2007. You can download the dataset at https://pan.baidu.com/s/12W0PsiwKIQj6AD3fx6j6MA and the password is ofkv.
If you find this dataset or code useful you can cite us using the following bibTex:
@Article{rs11091117,
AUTHOR = {Zhang, Haopeng and Deng, Qin},
TITLE = {Deep Learning Based Fossil-Fuel Power Plant Monitoring in High Resolution Remote Sensing Images: A Comparative Study},
JOURNAL = {Remote Sensing},
VOLUME = {11},
YEAR = {2019},
NUMBER = {9},
ARTICLE-NUMBER = {1117},
URL = {http://www.mdpi.com/2072-4292/11/9/1117},
ISSN = {2072-4292},
DOI = {10.3390/rs11091117}
}