Segment Anything Model for Road Network Graph Extraction
This work addresses the challenge of efficient large-scale road network extraction for applications like mapping and urban planning, though it is incremental as it builds on the existing SAM foundation.
The paper tackles the problem of extracting vectorized road network graphs from satellite imagery by adapting the Segment Anything Model (SAM) into SAM-Road, which directly predicts graph vertices and edges without complex post-processing, achieving comparable accuracy to the state-of-the-art method RNGDet++ while being 40 times faster on the City-scale dataset.
We propose SAM-Road, an adaptation of the Segment Anything Model (SAM) for extracting large-scale, vectorized road network graphs from satellite imagery. To predict graph geometry, we formulate it as a dense semantic segmentation task, leveraging the inherent strengths of SAM. The image encoder of SAM is fine-tuned to produce probability masks for roads and intersections, from which the graph vertices are extracted via simple non-maximum suppression. To predict graph topology, we designed a lightweight transformer-based graph neural network, which leverages the SAM image embeddings to estimate the edge existence probabilities between vertices. Our approach directly predicts the graph vertices and edges for large regions without expensive and complex post-processing heuristics, and is capable of building complete road network graphs spanning multiple square kilometers in a matter of seconds. With its simple, straightforward, and minimalist design, SAM-Road achieves comparable accuracy with the state-of-the-art method RNGDet++, while being 40 times faster on the City-scale dataset. We thus demonstrate the power of a foundational vision model when applied to a graph learning task. The code is available at https://github.com/htcr/sam_road.