Awesome

UDF4UFZ

This repository is the code of "A Unified Deep Learning Framework for Urban Functional Zone Extraction Based on Multi-source Heterogeneous Data"

Dependency

Pytorch
TorchVision
osgeo
numpy

Essential classes and function

models.py

Flowchart of the proposed unified deep learning framework for UFZ extraction The models are in file "models.py". There are two essential classes in the file. the complementary_fusion and the spatial_information_modeling. This two classes is corresponding to two parts in our paper.

distance_heatmap.py

Illustration of converting POIs into a hierarchical distance heatmap tensor The file contains the function of creating distance heatmap (creat_distance_heatmap). The inputs are

a raster imagery with geoinformation, which implies the processing extend,
a file contains POIs,
the output root, and
a zooming scale (default 10). You can also create the distance heatmap by ArcGIS (in ArcToolbox/Spatial Analyst Tools/Distance/Euclidean Distance)

If you need any technical support, never hesitate to contact us via salvial@csu.edu.cn.