基于Dlib与YOLO11改进的驾驶员疲劳分心检测及预警系统

打开文本图片集

中图分类号：TP391.41 文献标志码：A 文章编号：2095-2945（2026）01-0055-04

Abstract：Therewillbe facial oclusionscenes whenthedriver isactuallydriving，suchaswearing glases，wearing amask， etc.ThetraditionalDlibalgorithmthatonlyextractsthedriversfacialfeaturesforfatiguedetectionisnolongerapplicable.This papercombinesDlibandYOLOl1tousemulti-thresholddeterminationtoimprovethetraditionalDlibfatiguedetectionalgorithm. It providesafatiguedetectionalgorithmfordriverswearingglasses，masksandotherfacialoccusionscenarios.Italsousespublic datasetsontheRaspberryPi5hardwareplatformtoverifytheauracyoftheimprovedalgorithmfordriverfatiguedetection.In adition，theimprovedalgorithmcanalsodetectandvoiceremindersofdistracteddrivingbehaviorssuchassmokingandphone calls，achieving more comprehensive detection and early warning of fatigue and distracted behaviors.

Keywords：fatigue driving;Dlib;YOLO11;Raspberry Pi5 ；multi-threshold determination

疲劳和分心驾驶作为引发交通事故的重大因素之一，严重影响驾驶员以及乘客的人身与财产安全，于是驾驶员疲劳分心检测及预警系统的研发成为汽车行业的重要课题。（剩余5100字）