Automated Detecting and Placing Road Objects from Street-level Images

摘要: Navigation services utilized by autonomous vehicles or ordinary users require the availability of detailed information about road-related objects and their geolocations, especially at road intersections. However, these intersections are mainly represented as point elements without information, even not available in current versions crowdsourced mapping databases including OpenStreetMap(OSM). This study develops an approach to automatically detect place them right location from street-level images. Our processing pipeline relies on two convolutional neural networks: first segments images, while second detects classifies specific objects. Moreover, locate detected objects, we establish attributed topological binary tree(ATBT) based urban grammar for each image depict coherent relations topologies, attributes semantics Then ATBT is further matched with map features OSM determine placed location. The proposed method has been applied a case Berlin, Germany. We validate effectiveness our object classes: traffic signs lights. Experimental results demonstrate that provides near-precise localization terms completeness positional accuracy. Among many potential applications, output may be combined other sources data guide