Building Detection in Airports through Remote Sensing Image Using YOLOv3 with Jet Saliency Map

Building detection system through the remote sensing of images has been widely studied. In this paper, we propose a model for the detection of buildings at airports through different levels of remote sensing image. The proposed model is improved using from the You Only Look Once (YOLO) algorithm, which is based on the idea of the convolutional neural network (CNN). We also adjust each input image to our model using the Jet Saliency Map. The buildings to be detected in this study are the passenger terminals, the control towers, the cargo buildings, and the hangars, for which the data set have been collected from 322 different airports in Asia including 4,933 images and 13,103 buildings. Furthermore, our improved model is also examined for efficiency and accuracy. The results show that it can detect the intended objects efficiently 85% upwards, which provides higher accuracy than the original model.