DeepLocBIM: Learning Indoor Area Localization Guided by Digital Building Models

Fingerprinting-based indoor localization is a cost-effective approach to provide coarse-grained location information for pedestrian mass-market applications without the requirement of installing additional positioning infrastructure. While most solutions aim at pinpointing exact user, estimating zone/area promising achieve more reliable prediction. Area predominantly utilizes predetermined building model segmentation obtain labels collected fingerprints. We propose novel multifloor area by directly predicting polygon zones that contain position user. Our learns construct from wall segments and thus predicted areas have high conformity underlying (semantic expressiveness). On self-collected as well on public fingerprinting data set, we compare our with two reference approaches. demonstrate utilized surface polygons are average up 50% smaller than those models semantic expressiveness requiring manual floor plan segmentation.