Current bird species identification mobile applications on the market are trained using prior state-of-the-art technology of Convolutional Neural Networks. As of recent, Vision Transformers (ViTs) have made much progress in vision classification tasks with performance matching and exceeding their counterparts. This project aims to explore how changing different parameters during the creation and training of a classifier powered by ViT technology affects the performance of the classifier in the task of bird species prediction of a given image. To showcase this a mobile application prototype (compatible with both iOS and Android devices) is created and tested.
PLEASE NOTE: You must be a member of the University of Lincoln to be able to view this dissertation. Please log in here.