Fast Object Detection For Remote Sensing Data Using a Mobile Device

With the rapid development of remote sensing techniques, the quantity and quality of remote sensing data have been greatly promoted. The automatic understanding and semantic analysis of satellite or aircraft remote sensing images are strongly required. In this paper, we design and implement a lightweight CNN based network for the detection task of remote sensing images analysis. Experimental results show that our model is robust to distinguish different object categories with high similarity. At the same time, we deployed the model to mobile devices, which maintain the detection accuracy and prediction speed.