A new interpretation technique of traffic signs, based on Deep Learning and Semantic Web

In this paper, we present a new robust interpretation and recognition technique of traffic signs, this work is part of a smart cities project, a solution for road traffic. Our technique is based on two technologies: Semantic Web technology to interpret a message sent by a traffic sign and Deep Learning technology to recognize its image. General strategy is composed of three main steps. First of these consists to implement traffic signs ontologies by one of Semantic Web languages. Second step consists to use SPARQL language queries to query these ontologies in order to recuperate necessary synonyms to interpret a message sent by a traffic sign. For final step consists to use Deep Learning technology to recognize different objects in traffic sign image. Multi-agents Systems (MAS) will be used as simulation environments with two agent categories: agents that represent traffic signs and agent that represent vehicles.