Web-based enrichment of bike sensor data for automatic geo-annotation

摘要: In this paper, we describe a multi-modal bike sensing setup for automatic geo-annotation of terrain types using web-based data enrichment. The proposed road/terrain classification system is mainly based on the analysis volunteered geographic information gathered by bikers. By participatory accelerometer and GPS sensor collected from cyclists' smartphones, which enriched with web services, able to distinguish between 6 different types. For data, employs random decision forest (RDF), compared favorably task against algorithms. classifies every instance road (over 5 seconds interval) maps results onto user coordinates. Finally, all instances, can annotate create more advanced route statistics. accuracy 92% 6-class 97% 2-class on-road/off-road classification.