This article was first published in CSCS 2022: Proceedings of the 6th ACM Computer Science in Cars Symposium.
We introduce the Synthetic Pedestrian Dataset (SynPeDS), which was designed to support a systematic safety analysis for pedestrian detection tasks in urban scenes. The dataset was generated synthetically with a real-time, physically based rendering pipeline, and it provides camera frames and, in part, associated LiDAR point clouds. It contains ground truth for semantic segmentation, instance segmentation, 2D and 3D bounding boxes, and, in part, pose information and body part segmentation. In particular, it comes with a large amount of meta information for in-depth performance and safety analysis (for example, addressing semantic properties of the pedestrians and their environment in the frames). Some scenarios were specifically designed to systematically cover certain safety-relevant or performance-reducing dimensions of the input space, defined in project KI Absicherung. The dataset does not claim to be complete or free of bias but aims to support coverage and data distribution studies.
