mmdetection3d

[CVPR2024] OneFormer3D: One Transformer for Unified Point Cloud Segmentation

[AAAI2025] UniDet3D: Multi-dataset Indoor 3D Object Detection

[CVPR 2024] Memory-based Adapters for Online 3D Scene Perception

[ICIP2023] TR3D: Towards Real-Time Indoor 3D Object Detection

[WACV2022] ImVoxelNet: Image to Voxels Projection for Monocular and Multi-View General-Purpose 3D Object Detection

[ECCV2022] FCAF3D: Fully Convolutional Anchor-Free 3D Object Detection

3D LiDAR Point Cloud based Bird's-Eye View (BEV) Object Detection framework using BEVFusion. Features inference, visualization, fine-tuning, and evaluation on nuScenes

mmdet3d-v0.17.1 demo，for learning and example

3D object detection on kitti and nuscenes multimodal LiDAR point cloud and multicamera data

Ground Plane Projection for Monocular General-Purpose 3D Object Detection

一个集成了 Vue 3 前端、Spring Boot 后端、YOLO 图像检测服务和 MMDet3D 点云检测服务的多模块项目。

Hybrid Attention and Convolutional Induction Fusion for LiDAR-Camera BEV perception.

Camera-radar sensor fusion for 3D object detection on nuScenes — comparing detection-level, mid-level, and early-fusion designs against camera-only and radar-only baselines.

Image to Bird's Eye View Projection for Monocular General-Purpose 3D Object Detection

2D/3D Tight Constraint for Monocular General-Purpose 3D Object Detection

A modular teacher-free self-distillation framework that enhances camera-only BEV 3D detectors through EMA-based spatial–temporal consistency and uncertainty-aware feedback without modifying inference-time architecture.

Full-stack pipeline that turns raw LiDAR point clouds into 3D object-detection and segmentation datasets with the real MMDetection3D engine — frames, annotations, and model checkpoints stored on Backblaze B2 over the S3-compatible API. For autonomous-vehicle and robotics teams; Next.js + FastAPI, B2 credentials only.

Multi-modal perception pipeline for roadwork zone detection using RGB camera and LiDAR point clouds — benchmarks 6 deep learning models across 3D object detection and semantic segmentation tasks.

Reproducible 3D LiDAR detection with TensorRT FP16, exact deterministic voxelization, and ROS 2.