Akademik Veri Yönetim Sistemi
31st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM SIGSPATIAL GIS), Hamburg, Almanya, 13 - 16 Kasım 2023, ss.125-128
Automatic building footprint detection from aerial LiDAR point cloud data is an important research area that has use in various research areas ranging from city modeling to disaster management. Various steps must be carried out to obtain building footprints represented as polygons. Specifically, Digital Surface Model (DSM) and Digital Terrain Model (DTM) should be estimated, and then different filters (e.g. adaptive threshold, morphological open/close) and indices (e.g. terrain ruggedness index) are used to remove noise and vegetation. However, there has been lack of open-source software to realize these steps. Consequently, computational reproducibility of the findings has remained limited. The 2022 ACM SigSpatial GIS Cup contributed to this issue by openly providing aerial LiDAR point cloud data for 20 test sites, and evaluated the submissions performance by relying on a modified Intersection over Union (IoU) metric. The aim of this demo paper is to describe the development of a computationally reproducible framework to realize the GIS Cup's objective. First, the developed Python package lasbuildseg is described, and then its use on five different test sites from the GIS Cup are evaluated.