Synthetic Urban Driving Dataset: RGB Frames and Pixel-Wise Semantic Labels Across Multiple Custom 3D Environments

A dataset release generated with CARLA across four custom synthetic environments. It includes RGB frames and pixel-wise semantic segmentation ground truth labels. This dataset was generated using CARLA 0.10.0 with default camera settings for both RGB and semantic segmentation. The underlying 3D environments were created with a custom framework detailed in the supplementary paper "3D Environment Generation from Sparse Inputs for Automated Driving Function Development" (https://doi.org/10.3390/vehicles8030047).