Traffic monitoring and analysis via street footage

• المؤلف الرئيسي: Chew, Darren Wen Cong
• مؤلفون آخرون: Ong Chin Ann
• التنسيق: Final Year Project
• اللغة: English
• منشور في: Nanyang Technological University 2024
• الموضوعات: Computer and Information Science; Image recognition; Traffic; Web development
• الوصول للمادة أونلاين: https://hdl.handle.net/10356/181296

This project implements the use of computer vision and machine learning technologies such as object detection and object segmentation. This component has been developed and designed to handle multiple challenges such as nighttime images, poor lighting, various camera angles, blurring due to car speeds and imbalanced data set. With this component, key metrics such vehicle count, and detection, clustering and clustering density can be acquired. Besides computer vision and machine learning, the project also implements an improvised clustering technique of DB scan to help detect and monitor density of the traffic. With this improvised component, users can monitor traffic conditions on each respective roads especially when there are multiple roads within a camera image. Once metrics of cluster density have been calculated by the improvised clustering technique, data is pipelined to an unstructured cloud database (Microsoft Azure Cosmo DB) using a cloud tool called Microsoft Azure Function. With this component, data can be transformed and reflected on a webpage using a python library Stream lit for further comprehensive analysis. The webpage component has been developed to use cases such as acquiring routing information from a starting point to various destinations, along with multiple visual representations such as graph plots to showcase fluctuations in traffic densities.