Repository logo
 

A Computer Vision Approach for Object Detection and Lane Segmentation in Autonomous Vehicles

dc.contributor.advisorSaadat Mehr, Aryan
dc.contributor.committeeMemberWahid, Khan
dc.contributor.committeeMemberWu, FangXiang
dc.creatorAhadi, Mahyar
dc.date.accessioned2024-09-13T18:02:32Z
dc.date.available2024-09-13T18:02:32Z
dc.date.copyright2024
dc.date.created2024-08
dc.date.issued2024-09-13
dc.date.submittedAugust 2024
dc.date.updated2024-09-13T18:02:33Z
dc.description.abstractThis thesis presents a comprehensive approach to enhancing road safety and traffic management through advanced object detection, classification, and segmentation techniques. The work is divided into three primary tasks: First, an improved YOLOv5 model is employed to detect and classify various objects commonly found on roads and streets, including vehicles (cars, trucks, buses), traffic lights, and pedestrians. This approach ensures accurate and efficient identification of critical elements in real-time scenarios. Second, a custom Convolutional Neural Network (CNN) is proposed for the detection and classification of traffic signs. The CNN architecture is specifically designed to identify and categorize traffic signs with high accuracy, ensuring robust performance across diverse environmental conditions. Finally, the thesis addresses the segmentation of road lanes under both normal and challenging weather conditions. This is achieved through the integration of CNN and Long Short-Term Memory (LSTM) networks. The combination leverages the spatial feature extraction capabilities of CNNs and the temporal sequence learning strengths of LSTMs, resulting in reliable lane segmentation even in adverse weather scenarios. The main challenges for this thesis include achieving high accuracy in object detection, classification, and segmentation under various real-world conditions. Ensuring robust performance of the YOLOv5 model for detecting and classifying road objects in dynamic environments, and maintaining accuracy of the custom CNN for traffic sign detection across different weather and visibility conditions, are significant challenges. Effective lane segmentation is also difficult in adverse weather due to reduced visibility and unpredictable surfaces. Balancing accuracy, computational efficiency, and real-time processing in these scenarios is a key challenge. The methodologies and models developed in this research contribute to the field of autonomous driving and intelligent transportation systems by providing improved accuracy and robustness in real-world applications. The integration of these advanced techniques aims to enhance the safety and efficiency of road networks, supporting the development of smarter and more reliable autonomous vehicles.
dc.format.mimetypeapplication/pdf
dc.identifier.urihttps://hdl.handle.net/10388/16004
dc.language.isoen
dc.subjectAutonomous Driving, Convolutional Neural Networks, Deep Learning, Lane Segmentation, LSTM, Object Detection, Traffic Sign Detection, UNet, YOLOv5
dc.titleA Computer Vision Approach for Object Detection and Lane Segmentation in Autonomous Vehicles
dc.typeThesis
dc.type.materialtext
thesis.degree.departmentElectrical and Computer Engineering
thesis.degree.disciplineElectrical Engineering
thesis.degree.grantorUniversity of Saskatchewan
thesis.degree.levelMasters
thesis.degree.nameMaster of Science (M.Sc.)

Files

Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
AHADI-THESIS-2024.pdf
Size:
5.38 MB
Format:
Adobe Portable Document Format
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
LICENSE.txt
Size:
2.27 KB
Format:
Plain Text
Description: