DEEP LEARNING-BASED TRAFFIC LIGHT DETECTOR

Nowadays, there are still drivers who habitually do not follow the traffic light rule; they do not stop at the red light. This dissertation presents a Deep Learning-based Traffic Light Detector. The proposed model performs traffic light detection using a Convolutional Neural Network (CNN) to e...

Full description

Main Author: ISMAIL, NUR HIDAYAH
Format: Final Year Project
Language: English
Institution: Universiti Teknologi Petronas
Record Id / ISBN-0: utp-utpedia.21747 /
Published: IRC 2020
Subjects:
Online Access: http://utpedia.utp.edu.my/21747/1/17007440_Nur%20Hidayah%20binti%20Ismail.pdf
http://utpedia.utp.edu.my/21747/
Tags: Add Tag
No Tags, Be the first to tag this record!
Summary: Nowadays, there are still drivers who habitually do not follow the traffic light rule; they do not stop at the red light. This dissertation presents a Deep Learning-based Traffic Light Detector. The proposed model performs traffic light detection using a Convolutional Neural Network (CNN) to extract specific color features. CNN consists of 6(six) Convolutional layers. It is a fully connected layer that takes the convolution or pooling output and determines the appropriate mark to identify the image. A survey has been carried out gauging the proposal's market potential; 37 respondents stated a need for a traffic light alert system. The system is developed on an Asus VivoBook 15 laptop. A webcam is used to capture the traffic light image. The output is in the form of audio that alerts the red color traffic light.