Paved2Paradise: A Cost-Effective and Scalable Approach for Generating Diverse and Realistic LiDAR Datasets
By deliberately collecting separate "background" and "object" datasets and intelligently combining them, Paved2Paradise can generate large, diverse, and realistic LiDAR datasets for training 3D object detection models in a cost-effective manner.