Real-time vehicle detection and tracking in video surveillance cameras

Abstract

This End of Degree Work is directed to the field of Artificial Intelligence, expressly it is focused on the study of the Deep Learning techniques. Particularly, convolutional neural networks are studied with the aim of developing a vehicle detection model that is able to recognize them in images and videos. YOLOV3 is the vehicle detection model that has been studied, trained and optimized all through this work, in order to develop a model capable of detecting cars, trucks and motors in real time. Finally, SORT tracking algorithm has been studied and joint with the YOLOV3 detection model into a final model which will be responsible of detecting and tracking vehicles all through different videos in real time.