基于嵌入式平台的人脸疲劳检测算法

提出一种基于眼部和嘴部相结合的人脸疲劳检测算法,利用深度学习对人脸特征点进行定位,并通过眼部特征点计算PERCLOS参数,计算过程中详细讨论了不同特征对于PERCLOS参数提取的精确性,最后用角度特征来计算PERCLOS参数,同时将PERCLOS的时序信号进行频域分析,进而检测眼部代表的疲劳程度。与眼部协同的嘴部也是疲劳的重要指标,详细论述并提出了一种眼部和嘴部的疲劳协同参数。实验结果表明本文提出的方法是有效的,可以定量反映人的疲劳程度。同时将其部署在移动端用于车载疲劳检测和使用手机过程中的疲劳检测。

Face Fatigue Detection Algorithm Based on Embedded Platform

A face fatigue detection algorithm based on the combination of eye and mouth is proposed. The depth feature is used to locate the facial feature points and PERCLOS parameters are calculated by the eye feature points. The different characteristics are discussed in detail in the PERCLOS parameters The accuracy of the extraction, the final use of angle characteristics to calculate PERCLOS parameters, while the PERCLOS timing signal frequency domain analysis, and then detect the degree of fatigue on behalf of the eye. The mouth associated with the eye is also an important indicator of fatigue, detailed discussion and put forward a eye and mouth fatigue coordination parameters. The experimental results show that the proposed method is effective and can quantitatively reflect the degree of human fatigue. While its deployment in the mobile side for vehicle fatigue testing and the use of mobile phone fatigue testing.