Create a 3D Cuboid with 3D Point Cloud files(Bin, Pcd) collected with LiDAR Sensor and with the 2D images. Process Sensor-Fusion project with calibration information. The precise and high-quality dataset can be created even in difficult sight identifying circumstances with 3D Point Cloud Object Detection model and Object Tracking model using deep learning. Labeling processes in Calibration, Sensor-Fusion, Meta-data, and Curation are done efficiently based on our abundant experiences accumulated from in-house operated autonomous vehicles to collect datasets for autonomous driving.