S3PT: Scene Semantics and Structure Guided Clustering to Boost Self-Supervised Pre-Training for Autonomous Driving
This work addresses challenges in autonomous driving by enhancing self-supervised pre-training to better handle rare classes and diverse object sizes, though it is incremental as it builds on existing clustering-based techniques.
The paper tackled the problem of imbalanced object classes and complex scene geometries in self-supervised pre-training for autonomous driving by proposing S3PT, a scene semantics and structure guided clustering method, which improved performance on downstream tasks like semantic segmentation and 3D object detection across datasets such as nuScenes, nuImages, and Cityscapes.
Recent self-supervised clustering-based pre-training techniques like DINO and Cribo have shown impressive results for downstream detection and segmentation tasks. However, real-world applications such as autonomous driving face challenges with imbalanced object class and size distributions and complex scene geometries. In this paper, we propose S3PT a novel scene semantics and structure guided clustering to provide more scene-consistent objectives for self-supervised training. Specifically, our contributions are threefold: First, we incorporate semantic distribution consistent clustering to encourage better representation of rare classes such as motorcycles or animals. Second, we introduce object diversity consistent spatial clustering, to handle imbalanced and diverse object sizes, ranging from large background areas to small objects such as pedestrians and traffic signs. Third, we propose a depth-guided spatial clustering to regularize learning based on geometric information of the scene, thus further refining region separation on the feature level. Our learned representations significantly improve performance in downstream semantic segmentation and 3D object detection tasks on the nuScenes, nuImages, and Cityscapes datasets and show promising domain translation properties.