Automatic Power Line Detection for Low-Altitude Aircraft Safety Based on Deep Learning

Abstract

Power line detection (PLD) is of vital importance for the flight security of low-altitude aircraft, such as helicopters and unmanned aerial vehicles (UAVs). This paper firstly summarises past studies on PLD based on image processing technique extensively. Secondly, different from the traditional PLD methods, we propose an approach to detect power lines based on deep learning which has been demonstrated having unparalleled performance in the field of image processing and computer vision. Specifically, the convolutional neural network (CNN) is employed in this study to extract features and thus detecting power lines from images. By utilising CNN, the feature extraction and object detection process are completed jointly unlike traditional PLD methods. A public dataset is used to demonstrate the performance of the proposed approach. This paper also gives recommendations for the future development of PLD.