Abstract. Although deep learning has greatly improved the semantic segmentation accuracy of point clouds, the segmentation of rare classes in large-scale urban scenes has not been targeted in available methods. This paper proposes a two-stage segmentation framework with automated workflows for imbalanced rare classes based on general semantic segmentation. The proposed approach includes two stages: general semantic segmentation and object-based refined semantic segmentation. Firstly, general segmentation networks are utilized to segment general large objects. Secondly, refined semantic segmentation is conducted by an automated workflow: 3D clustering and bounding box (BBox) generation are applied to the point cloud of rare fine-grained objects during the training, followed by object detection to extract fine-grained objects. Afterwards, as the constraints, the extracted BBoxes further refine the segmentation results. Our approach is evaluated on the Hessigheim High-Resolution 3D Point Cloud (H3D) Benchmark and obtains state-of-the-art 89.35% overall accuracy and outstanding 75.70% mean F1-Score. Furthermore, rare classes Vehicle and Chimney achieve breakthroughs from zero to 63.63% and 52.00% in F1-score, respectively.
Abstract. Urban areas are complex scenarios consisting of objects with various materials. This variety poses a challenge to single-data classification schemes. In this paper, we propose a feature fusion and classification network on RGB top-view point cloud and SAR images with swin-Transformer. In this network, the heterogeneous features are learned separately by an asymmetric encoder, and then they are concatenated along the channel dimension and fed into a fusing encoder. Finally, the fused features are decoded by an UperNet for generating the semantic labels. As data we use high-resolution 3D point cloud provided by Hessigheim benchmark which are complemented by TerraSAR-X images. The overall precision and the mean intersection over union (mIoU) achieves 87.25% and 73.56%, respectively, which outperforms the single-data swin-Transformer by 4.08% and 1.91%, respectively.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.