This repository provides the official Python implementation of AB3DMOT, a real-time 3D multi-object tracking framework published at IROS 2020. The system tracks objects like cars, pedestrians, and cyclists using 3D bounding boxes from sensors like LiDAR, with support for standard autonomous driving datasets including KITTI and nuScenes. It also introduces new evaluation metrics specifically designed for 3D tracking to better assess tracking accuracy in robotics and autonomous vehicle applications.