We present "DetecTree" an open-source library that automatically detects urban trees from aerial imagery based on an existing pixel-level classification procedure. The library is based on the scientific Python stack and is organized as a reusable package that it can be easily installed and integrated into complex computational workflows. An example repository with reproducible Jupyter notebooks shows how "DetecTree" can be used to detect trees in the city of Zürich. Additionally, we show how "DetecTree" is used in an academic work to study the relationship between land use/land cover patterns and urban heat islands.

Download the slides for this talk.Download ( PDF, 2308.78 MB)