A tour guide implementation based on different crowd levels

Abstract

This dissertation aims to deliver a system where a user is guided through an environment, in such a way that their experience does not interfere with that of others where possible. The system is built using a client-server architecture where the server monitors current crowd density and future congestion to a certain degree. The clients are the users who are equipped with an Android application which communicates with the server to build a path according to the users' wishes. The server uses Histogram of Oriented Gradients, developed by Dalal and Triggs [1], to detect human figures and build a topology of the area using this information. The server estimates future congestion by receiving information from the clients, informing it of what their next destination is going to be. The client system is an Android application which acts as a tour guide for the user and guides them through the area. Users are able to request their next destination, prioritising factors which affect their experience. A web service is also made available for clients to be able to communicate with the server and retrieve information about the current state of the area being monitored. A thorough evaluation of the system is presented. A set of tests were carried out in order to demonstrate the performance of both aspects of the system. The tests carried out on the two HOG detectors implemented returned more than acceptable results. Those carried out on the tour construction algorithm also proved satisfactory.