A XGBoost Algorithm-based Fatigue Recognition Model Using Face Detection

TL;DR

This paper presents a fatigue recognition model based on XGBoost that uses facial indicators EAR and MAR, achieving high accuracy and sensitivity, suitable for practical applications.

Contribution

It introduces a novel fatigue detection model combining face detection with XGBoost using EAR and MAR indicators, demonstrating improved performance.

Findings

Accuracy of 87.37%

Sensitivity of 89.14%

Effective for fatigue detection

Abstract

As fatigue is normally revealed in the eyes and mouth of a person's face, this paper tried to construct a XGBoost Algorithm-Based fatigue recognition model using the two indicators, EAR (Eye Aspect Ratio) and MAR(Mouth Aspect Ratio). With an accuracy rate of 87.37% and sensitivity rate of 89.14%, the model was proved to be efficient and valid for further applications.