Image-Conditioned Graph Generation for Road Network Extraction
This work provides a proof of concept for applying graph generation models to road network extraction, addressing a domain-specific challenge in geospatial analysis.
The authors tackled the problem of road network extraction from image data by developing the Generative Graph Transformer (GGT), an autoregressive model that uses attention mechanisms for image-conditioned graph generation, and introduced the Toulouse Road Network dataset and StreetMover distance for benchmarking.
Deep generative models for graphs have shown great promise in the area of drug design, but have so far found little application beyond generating graph-structured molecules. In this work, we demonstrate a proof of concept for the challenging task of road network extraction from image data. This task can be framed as image-conditioned graph generation, for which we develop the Generative Graph Transformer (GGT), a deep autoregressive model that makes use of attention mechanisms for image conditioning and the recurrent generation of graphs. We benchmark GGT on the application of road network extraction from semantic segmentation data. For this, we introduce the Toulouse Road Network dataset, based on real-world publicly-available data. We further propose the StreetMover distance: a metric based on the Sinkhorn distance for effectively evaluating the quality of road network generation. The code and dataset are publicly available.