Deep-Learning Based Image Classification for Autonomous Vehicles

Abstract

Image classification is no doubt implemented within many applications and systems and continues to be one of the leading classical problems within the machine learning, computer vision and image processing fields. Autonomous driving vehicles (AV’s) utilise a supervised learning approach to develop leading systems to assist vehicles to carry out on-road tasks as safely as possible. Unfortunately, road collision is an ongoing major global issue, which does not only have global economic impact, but also has social implications on the lives of devastated families and friends who have lost loved ones in on-road car accidents. This project aims to explore and highlight the critical issues related to autonomous driving vehicles, image classification, Convolutional Neural Networks (CNN’s) and to build an image classification model that successfully classifies unseen images of two different objects (Cars, Dogs). The project aims to also demonstrate the importance of computer vision concerning autonomous driving vehicles and how utilising image classification with respect to AV’s could aid in minimising on-road collisions.