City2Graph Library Released for GeoAI

City2Graph is a Python library that converts geospatial data into analyzable graph formats for spatial analysis, network analysis, and GeoAI (Graph Neural Networks).

Key features include:

• Morphology: Uses OpenStreetMap and Overture Maps to generate graphs of buildings, roads, and urban structures
• Transportation: Loads GTFS and GBFS data to build transportation graphs between stops
• Mobility: Processes OD matrices and flow data (migration, bike sharing, pedestrian counts, etc.) as weighted spatial graphs
• Proximity: Supports KNN, Delaunay, Queen/Rook adjacency, etc., based on Euclidean, Manhattan, and network distances
• Heterogeneous Graphs: Includes various node and edge types, with relationships constructed via metapaths

This library supports bidirectional conversion between GeoDataFrame, NetworkX, and PyTorch Geometric (HeteroData), preserving geometric structures and attributes during the conversion process.