OpenPCDet is a deep learning toolbox providing state-of-the-art 3D object detection models for LiDAR point clouds. It implements multiple neural architectures including PointRCNN, PV-RCNN, Voxel R-CNN, BEVFusion, and DSVT, supporting multi-modal detection on datasets like Waymo, NuScenes, and Argoverse2. The project includes model training, evaluation, and inference pipelines with TensorRT acceleration support.