Helmet Wearing Detection System for Non-Motor Vehicle Riders Based on k210 and YOLOv3

Abstract

Wearing helmets can effectively reduce the secondary injuries caused by traffic accidents, and it is necessary to detect the helmet wearing conditions of non-motor vehicle riders. This paper proposes a portable and low-cost rider helmet wearing detection system. The system adopts the lightweight network framework of YOLOv3, and deploys the algorithm on the K210 microcontroller. The innovation of this paper is that the calculation amount and model size of the traditional YOLOv3 algorithm model are much larger than the maximum scale supported by the K210 chip, and it is difficult to deploy on small embedded devices. We need to improve the traditional YOLOv3 algorithm. This paper adopts the method of replacing the backbone network to reduce the complexity of the model, so that the algorithm can be deployed on the K210 microcontroller. This paper conducts comparative training under two different backbone networks, and deploys the appropriate model on the k210 single-chip microcomputer, which reduces the detection cost.