Road Network and Travel Time Extraction from Multiple Look Angles with Spacenet Data

摘要

Identification of road networks and optimal routes directly from remote sensing is of critical importance to a broad array of humanitarian and commercial applications. Yet while identification of road pixels has been attempted before, estimation of route travel times from overhead imagery remains a novel problem, particularly for off-nadir overhead imagery. To this end, we extract road networks with travel time estimates from the SpaceNet MVOI dataset. Utilizing the CRESIv2 framework, we demonstrate the ability to extract road networks in various observation angles and quantify performance at 27 unique nadir angles with the graph-theoretic APLSlength and APLStime metrics. A minimal gap of 0.03 between APLSlength and APLStime scores indicates that our approach yields speed limits and travel times with very high fidelity. We investigate the performance of angle-specific models, as well as the utility of incorporating all available angles during model training, yielding a peak score of APLStime = 0.56. The combined model exhibits greatly improved robustness over angle-specific models, despite the very different appearance of road networks at extremely oblique off-nadir angles versus images captured from directly overhead.