Semantic segmentation on Swiss3DCities: A benchmark study on aerial photogrammetric 3D pointcloud dataset

We introduce a new outdoor urban 3D pointcloud dataset, covering a total area of 2.7km(2), sampled from three Swiss cities with different characteristics. The dataset is manually annotated for semantic segmentation with per-point labels, and is built using photogrammetry from images acquired by multirotors equipped with high-resolution cameras. In contrast to datasets acquired with ground LiDAR sensors, the resulting point clouds are uniformly dense and complete, and are useful to disparate applications, including autonomous driving, gaming and smart city planning. As a benchmark, we report quantitative results of PointNet++, an established point-based deep 3D semantic segmentation model; on this model, we additionally study the impact of using different cities for model generalization. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

The dataset introduced is an outdoor urban 3D pointcloud dataset covering three Swiss cities with different characteristics and a total area of 2.7km(2). This dataset, manually annotated for semantic segmentation, is suitable for various applications such as autonomous driving, gaming, and smart city planning. The quantitative results of PointNet++, a point-based deep 3D semantic segmentation model, are reported as a benchmark, with additional study on the impact of using different cities for model generalization.