Implementation of Yolov5-Based Camera on Bicycle for Collision Prevention

Cycling is essential for many people as a means of transportation, exercise, recreation, and deliveries, offering benefits for both individuals and the environment. However, cyclists are more vulnerable to collisions and have limited rear visibility. This study explores using the YOLOv5 algorithm to detect oncoming traffic behind cyclists, assisting their awareness. The system consists of a single USB camera attached to a Raspberry Pi Model 4B and is designed to detect and classify common vehicles found in Metro Manila roads. The system is then attached to a bicycle for the training and testing of the system, which were conducted inside North Belton, a subdivision in Quezon City, where results yielded a Mean Average Precision (mAP) of above 0.8.