From Pixels to Progress: Generating Road Network from Satellite Imagery for Socioeconomic Insights in Impoverished Areas

TL;DR

The paper tackles the problem of incomplete road data in impoverished regions, proposing a scalable pipeline that combines data filtering, a connectivity-aware encoder-decoder road segmentation model, morphological processing, and a graph-extraction step to generate road networks from satellite imagery. It introduces a comprehensive road-network dataset covering 382 impoverished counties in China (794,178 km^2, 17.048 million people) and demonstrates substantial improvements in topology and geometry, including a 42.7% rise in F1-score and ~80% road reconstruction relative to baselines. The work couples this technical framework with socioeconomic analysis, showing scaling laws between population and road length and a positive, measurable impact of road expansion on GDP via Difference-in-Differences. This combination provides a practical tool for SDG monitoring and LNOB policy planning in data-poor regions, with broad applicability to infrastructure and development analytics.

Abstract

The Sustainable Development Goals (SDGs) aim to resolve societal challenges, such as eradicating poverty and improving the lives of vulnerable populations in impoverished areas. Those areas rely on road infrastructure construction to promote accessibility and economic development. Although publicly available data like OpenStreetMap is available to monitor road status, data completeness in impoverished areas is limited. Meanwhile, the development of deep learning techniques and satellite imagery shows excellent potential for earth monitoring. To tackle the challenge of road network assessment in impoverished areas, we develop a systematic road extraction framework combining an encoder-decoder architecture and morphological operations on satellite imagery, offering an integrated workflow for interdisciplinary researchers. Extensive experiments of road network extraction on real-world data in impoverished regions achieve a 42.7% enhancement in the F1-score over the baseline methods and reconstruct about 80% of the actual roads. We also propose a comprehensive road network dataset covering approximately 794,178 km2 area and 17.048 million people in 382 impoverished counties in China. The generated dataset is further utilized to conduct socioeconomic analysis in impoverished counties, showing that road network construction positively impacts regional economic development. The technical appendix, code, and generated dataset can be found at https://github.com/tsinghua-fib-lab/Road_network_extraction_impoverished_counties.

Figures (13)

• Figure 1: Comparison of (a) one satellite image, (b) corresponding OSM road network, and (c) our generated road network. (Latitude=$33.6945^\circ$N, Longitude=$110.3237^\circ$E). The roads in red rectangles in figure (c) are missing in OSM data.
• Figure 2: The overall framework of the proposed road network extraction framework.
• Figure 3: Visualization of the road network extraction results in two impoverished counties. The roads are plotted in red. The locations of the two counties are (Latitude=$39.4255^\circ$N, Longitude=$114.2517^\circ$E) and (Latitude=$32.9858^\circ$N, Longitude=$107.7386^\circ$E).
• Figure 4: The results of reconstructing different classes of roads in $2021$.
• Figure 5: The spatial distribution of selected impoverished counties. Different colors mark different economic regions and the circle size shows the relative road length in $2021$.