Proposal for a Sign Detection Algorithm for an Autonomous Vehicle

Abstract

Vehicle automation is a field that is gaining an increasing amount of traction, and has seen much success recently. Yet one nagging issue still remains: sign detection. Despite the wide range of research being done into autonomous vehicles, no vision algorithm that can identify all signs in real time has been developed. This thesis uses two methods, a state of the art object detector, and a simple contour finder, to attempt to handle this issue. However, the lack of progress in this area calls into question whether computer vision is the best way to approach this problem.