Classification Between Pedestrians and Motorcycles using FasterRCNN Inception and SSD MobileNetv2

Abstract

Philippines is on the list of the top ten fastest growing economy in the world. One of its developments is in its traffic law enforcement. Today, the government is continuously finding ways on how to alleviate the problem on its roads with the use of technology. The Metro Manila Development Authority (MMDA) has offered a solution which is the No Contact Traffic Apprehension Policy, that utilizes Closed-Circuit Television (CCTV) Monitoring to apprehend vehicles violating traffic laws, rules and regulations by capturing videos and images. To further enhance this, the government has partnered with the De La Salle University to use artificial intelligence in the system with the project “CATCH-ALL”. With the use of CCTVs and artificial intelligence system, it can help the system detect traffic violations using an automated process. But some tasks are not that easy to execute by the computer like differentiating a pedestrian and a motorcycle. In this study, a novel approach to classifying a pedestrian and a motorcycle with the use of object detection will be developed. It will be demonstrated using deep machine learning, specifically convolutional neural network and by utilizing different pre-trained models to a gathered dataset.