Design and development of a computer-assisted Road Physical Feature Extraction (RFEX)

• Main Authors: Agulto, Juliana Marie Balay, Benabaye, Eric Montecastro, Gutierrez, Gian Carlo Dela Cruz, Noroña, Yeohan Lorenzo M.
• Format: text
• Language: English
• Published: Animo Repository 2023
• Subjects: Highway engineering > Computer programs; Roads—Design and construction > Computer programs; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/etdb_comtech/10
• Institution: De La Salle University

The concept of livability plays a crucial role in urban planning, development, and maintenance. Analyzing and interpreting physical road features is essential for assessing the livability of an area. However, there is currently no existing database available that contains the road physical features of Metro Manila. To address this gap, the proposed Computer-Assisted Road Physical Feature Extraction System (RFEX) aims to create a database of road characteristics. The system focuses on detecting road physical features, including the presence of road lanes, bike lanes, sidewalks, fences, obstructions and constructions. The system incorporates modules for lane detection and object detection, along with a user-friendly website. The core object detection of the system achieved an average precision, recall, and mAP of 0.72, 0.46, and 0.52, respectively. For both bike lane detection and road lane counting, the average accuracy achieved 0.64, and the average mean absolute error achieved 0.37, respectively, by utilizing the unit factor k=1 in median absolute deviation. Our proposed work provided a computer vision approach to detecting bike lines and counting road lanes as well as developing a system capable of storing and visualizing the extracted road physical features. By establishing this, people can gain valuable insights regarding road infrastructures.