This repository provides a PyTorch implementation of Complex-YOLO, a deep learning architecture derived from YOLOv4 designed for 3D object detection using LiDAR point clouds. It supports distributed data parallel training, uses GIoU loss for rotated bounding box optimization, and includes mosaic/cutout data augmentation. The model is intended for autonomous driving applications where it detects objects like vehicles and pedestrians from 3D sensor data.