Investigating CNN-based YOLO model architectures to improve vehicle detection accuracy in intelligent transportation systems

Sudharsan, Gowtham (2025) Investigating CNN-based YOLO model architectures to improve vehicle detection accuracy in intelligent transportation systems. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (379kB) | Preview

Abstract

Growing vehicle density in urban areas establishes major challenges to modern traffic management, particularly during harsh weather conditions. This paper investigates the potential of CNN-based YOLO models, i.e., YOLOv7, in achieving enhanced accuracy for detecting vehicles under intelligent transportation systems. Even though YOLOv7 has been most renowned for its real-time detection speed and detection accuracy, research has excluded it from its vulnerabilities in poor weather conditions. To connect this failing, the paper integrates YOLOv7 object detection with a ResNet50 model trained on the DAWN dataset to identify the state of weather by training YOLOv7 on the BDD10K dataset for the detection of eight traffic object classes where the models were coded in PyTorch and validated using the standard measurements of mAP, precision, recall, and classification accuracy. Results demonstrate the end-to-end pipeline’s high accuracy across various environmental conditions with mAP@0.5 of 60.5% for object detection overall and over 85% for weather classification. A real-time dashboard simulator was developed with Gradio, illustrating the way in which the system can potentially be integrated into smart city systems. This two-model solution offers an end-to-end, scalable, modular, and context-aware approach to enhancing situational awareness, congestion control, and autonomous vehicle decision-making in high-density urban scenarios.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Kelly, John UNSPECIFIED
Subjects:	H Social Sciences > HE Transportation and Communications Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence > Computer vision Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence > Computer vision H Social Sciences > HE Transportation and Communications > Urban Transportation
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	26 Aug 2026 12:19
Last Modified:	26 Aug 2026 12:19
URI:	https://norma.ncirl.ie/id/eprint/9671

Actions (login required)

View Item