Tracking of drone flight by neural network Siamese-RPN

Abstract

Recently deep learning has made a great progress in the image classification and video object tracking. However the speed of traditional region proposal algorithms is too slow and become a bottleneck. Researchers have proposed a lot of neural networks for the improvements. With the region proposal network proposed, neural network can finish the computation enabling cost-free region proposals, which make it possible to do real-time detection and tracking. This paper will set up a system which can detect and track drone by using neural networks such as Faster R-CNN and Siamese-RPN.