U-Net Based Moving Target Detection for Unmanned Air Vehicles
Qingtao Yu, Haodi Yao, Fenghua He, Hongsheng Qi · 2018
In this paper, a moving target detection problem is investigated for the unmanned air vehicles with a monocular vision sensor. First, a detection algorithm based on U-Net network is proposed to obtain the position and velocity direction of a moving target. Then, a Kalman Filter is incorporated into the proposed algorithm to increase the accuracy of states acquisition of the moving targets. Then, the experimental results show the proposed algorithm has low computation and better performance.