Faculties and centres
Automatic Understanding and mapping of outdoors regions in cities using Google Street View images
Dr. José Carlos Rangel, investigador en la Universidad Tecnología de Panamá
Abstract: The use of semantic representations to achieve place understanding has been widely studied using indoors information. Semantic descriptors allow to get an idea of the role of a place, these descriptors are usually generated with a deep neural network. Then, this kind of data can be used for navigation, localization and place identification using mobile devices. Nevertheless, applying this approach to outdoor data involves some non-trivial procedures such as the gathering of the information. This problem can be solved using maps APIs such as Google Street View that allows getting images from the dataset captured for adding to the map of a city. In this paper, we seek to make the most of this kind of APIs for collecting images of the street of a city. Then, generating a semantic representation of the city built using a clustering algorithm and semantic descriptors. The proposed method can automatically assign a semantic label for the cluster on the map. Experimental results were carried on using several clustering distances for getting plenty of different maps that were analyzed for evaluating the proposal. Results show the goodness of the use of Google Street View images, semantic descriptor, supervised and unsupervised machine learning algorithms for generating semantic maps for external places. These maps properly encode the zones existing in the selected city.