Target tracking is a hot topic in the field of computer vision and has a wide range of applications in practice.The 3D single-target tracking problem is mainly studied in this paper.Considering that the 3D target detection has good performance, the 3D target detection result obtained by using the frustum point cloud network is combined with the TLD target tracking algorithm, and the detection module of the TLD algorithm is improved.A 3D single-target tracking algorithm based on the TLD algorithm is proposed.The input is the RGB-D image sequence, and the output is the 3D tracking results of the target.The test results in the KITTI data set prove the feasibility of the algorithm in 3D target tracking.