Comparison of artificial intelligence and traditional image processing algorithms for intelligent transportion systems

• Main Author: Manushri, Akunuri
• Other Authors: Mohammed Yakoob Siyal
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering
• Online Access: https://hdl.handle.net/10356/176752
• Institution: Nanyang Technological University

The project aims to compare Artificial Intelligence and traditional image processing algorithms and analyse which is a better fit for road traffic analysis. Maximising the effectiveness and capacity of any contemporary transport is essential when it comes to road traffic data. The project hopes to understand the effectiveness of the different ways by gathering the statistical data for traffic analysis. Several established traffic monitoring techniques were researched and implemented, including edge detection, background difference, and inter-frame differencing. Traffic video samples from city roads were collected under various lighting conditions. These samples were then extracted into frames and analysed using different image processing techniques within Python and AI approaches, specifically the YOLOv8 Convolutional Neural Network model, were applied to obtain quantitative data on vehicle classification and count. This allowed for a comparative analysis to identify superior algorithms and techniques. The results demonstrated the superiority of using Artificial Intelligence for traffic analysis. Additionally, the project compared front and back angle video footage to determine the optimal perspective, along with analysing performance under different lighting conditions to identify any alignment in results.