|UAVs Detection and Tracking by Features-Enhanced Yolo and Kalman Filtering
|Unmanned Aerial Vehicles defense becomes a hot topic due to the ongoing Ukraine-Russia War and China’s constant threats to Taiwan. Drones tracking and countermeasures inevitably play a crucial role in future warfare. In this talk, we propose a feature-enhanced Yolo network to improve the accuracy of small moving objects detection. Furthermore, to achieve real-time tracking, Kalman filter (KF) is applied to reduce the search window. A structure-preserving algorithm is proposed to estimate the covariance in KF. Furthermore, FPGA is employed to accelerating the computation of the covariance estimation. Experimental results will be shownto demonstrate the effectiveness of our approach.