Aiming at the problems of difficult detection and easy penetration of traditional radar detection equipment when facing low-speed-small UAV, an UAV recognition and trajectory prediction algorithm based on deep convolution neural network and long-short-term memory neural network is proposed. The deep convolution neural network is used to identify the UAV in real time and extract the target category and image space position information. According to the change trend of UAV image space position in time domain, the UAV flight mapping trajectory is drawn. The long-short-term memory network is used to predict the flight mapping trajectory, and the predicted track direction of UAV in the future time domain is obtained to realize the unmanned aerial vehicle early warning and tracking, real-time detection and trajectory inference.The experimental results show that the average accuracy of target recognition and trajectory prediction can reach 82% and 80% in 24 frames per second. It can give real-time and accurate warning to UAV in the ground-based computing platform, and can effectively prevent the penetration and penetration of non-cooperative UAV in the airspace.