Anyue YuShuxi ChenLi ZhangBaoxiang Qian0000-0003-0793-3308Chin Soon Ku2026-11-062026-11-062025-08-28978164368609710.3233/ATDE250693https://dspace-cris.utar.edu.my/handle/123456789/11857Against the backdrop of economic changes that have increased employees' work pressure and drowsiness issues, and in the context of a shortage of employee drowsiness monitoring systems in enterprises, this study designed and implemented an employee drowsiness monitoring system based on computer vision. This system utilizes the pre - trained models of OpenCV and Dlib libraries to determine employees' drowsy states by detecting facial key points, blinks, yawns, and nods. The research recruited 10 employees for 500 independent tests, each lasting 10 minutes, covering various scenarios. The results show that the average accuracy rate of this system reaches 98%, far exceeding that of the CNN - LSTM based system. Moreover, it has low requirements for computing resources, low deployment costs, and a short time to be put into use. Evidently, this system effectively addresses the problem of employee drowsiness monitoring in enterprises and has a promising application prospect. © 2025 The Authors.en-UScomputer visioncomputing resourcesdetection accuracyDlib LibraryEmployee drowsiness monitoring systemOpenCVArtificial intelligenceEmploymentInformation systemsInformation usePersonnel10 minutesComputing resourceDesign and implementationsDetection accuracyDlib libraryKeypointsMonitoring systemOpencvWork pressuresComputer visionDesign and Implementation of an Employee Drowsiness Monitoring System Based on Computer Visiontext::conference output::conference proceedings::conference paper