Facial biometrics system with YOLOv8-Deep Learning to improve the user experience in public transport: Case study in Metropolitan Lima, Peru-2024 (#604)

Authors

Yopan Cuipal, Juber Antony

Quispe Llacctarimay, Ruben

Abstract

Urban bus transport in Metropolitan Lima is a problem that the authorities have not been able to control, causing citizens to lose time, exposing them to danger and a terrible experience. Therefore, it is essential to develop a solution and design it to improve the service that companies offer to the user. The study seeks to design the solution through facial recognition in order to improve the user experience using YOLOv8-Deep Learning on public transport buses operating in Metropolitan Lima. For this purpose, the area, type of bus, the ideal way to reduce processing resources and to make it as interactive as possible for the drivers of the units were analysed. The detection area was defined and a pre-trained Yolov8 model with counting was used to compare this value with the maximum capacity entered as the key to activate or deactivate the bus boarding and alighting access doors with other indicators. The test results were acceptable, obtaining, with optimal lighting, about 90% accuracy, managing to control and count the number of passengers in the bus passageway, however, there were limitations related mainly to the lighting that, if it exceeded 60% of absence of light, regardless of the resolution of the cameras, the accuracy had a drop to less than 20%. The project will improve the quality of service and avoid the bad driving tactics employed, which will reduce accidents, reduce hours of congestion and provide a continuous flow of traffic in a chaotic city like Metropolitan Lima.