Detekce motorových vozidel ve 2D obraze

Abstract

Computer vision, image processing and machine learning belong to the most dynamically developing branches of modern computer science. Their goal is to teach computers how to comprehend and distinguish real world objects without any simplifications respectively teach them to evaluate inputs from real world and react individually or execute series of logical steps without explicit interference of people. Such approach has wide use in almost all branches of human work beginning with simple image processing through automatic text recognition ending with automatic evaluation of medical pictures and specifying diagnoses. This diploma thesis focuses mainly on usage of mentioned techniques in traffic adjustment and autonomous driving systems. Primary goal of final application is to recognize road, other vehicles and obstacles on road in images received from camera mounted in the front part of the vehicle which may lead to higher safety and fluency of traffic.