Motorcycle and Vehicle Detection for Applications in Road Safety and Traffic Monitoring Systems

• Main Authors: Cruz, Gerson Gerard L, Litonjua, Aaron, San Juan, Alysia Noreen P, Libatique, Nathaniel Joseph C, Tan, Marion Ivan L, Honrado, Jaime Luis E
• Format: text
• Published: Archīum Ateneo 2022
• Subjects: monitoring systems; object detection; object tracking and counting; OpenCV; road safety; YOLOv4; Computer Sciences; Electrical and Computer Engineering; Engineering; Other Electrical and Computer Engineering; Systems and Communications
• Online Access: https://archium.ateneo.edu/ecce-faculty-pubs/130; https://doi.org/10.1109/GHTC55712.2022.9910992
• Institution: Ateneo De Manila University

Motorcycles are becoming increasingly common in middle to low-income countries as cheaper alternatives to fourwheeled vehicles. The reliance on motorcycle-based services has also seen a substantial increase in popularity, leading to a greater proportion of motorcycles on the road. The increase in motorcycle reliance necessitates a need for motorcycle-inclusive road information generation as motorcycles are the most susceptible to fatal road crashes. We report the results of our application of the You Only Look Once (YOLOv4) algorithm to count and classify vehicles and motorcycles in traffic videos obtained by our group over a three-month period along Katipunan Avenue Southbound (KAS), Metro Manila. This has been made to run in real-time with video and is able to process a video output with its annotations and a counter for both classes. These results show that a motorcycle and vehicle detection and counting system can be feasibly considered for data-driven road safety and traffic monitoring systems.