Intelligent Mobile-Based Campus Navigational Assistant Using Natural Language Processing and Computer Vision: A Case Study for University of Ghana

Navigating through the University of Ghana campus, like most tertiary campuses, can be very challenging, especially for a freshman, foreign, or an exchange student. There are numerous routes leading to particular buildings, and depending on the route taken, one may not see the inscriptions placed on these buildings. Some buildings do not have any inscription at all, and some of the signboards to the buildings have defaced due to prolonged exposure to harsh environmental conditions. Smartphones are ubiquitous today, and the current trend of conversations among the youth is dominant through mobile chat. Therefore, this project sought to take advantage of these factors to design a mobile-based system to cater to this need by providing a means of building identification and providing map routes to help students find their way around. Neural Networks was employed in the development of the project. Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are two mainstream methods of deep learning and as such, were incorporated to develop a mobile-based chatbot system (UGBot) that provides an interactive manner of helping students identify buildings by taking pictures of the building and requesting for directions to their intended destinations. After implementing the classifier modules, accuracy values of 96% and 90% were obtained for the image and text classifiers, respectively.