基于机器学习的多视觉特征融合疲劳检测
首发时间:2016-01-12
摘要:疲劳检测对于降低交通事故发生率,保障出行安全,具有重要的意义。论文针对现有疲劳检测算法特征少,稳定性差及准确度低等弊端,提出了一种基于机器学习的多视觉特征融合的疲劳检测算法。首先,论文综合考虑包括眨眼频率、闭眼时长、打哈欠的频率、打哈欠的时长、点头的频率、点头的时长、摇头的频率以及摇头的时长在内的8个视觉疲劳特征,并用机器学习模型训练和学习疲劳特征,优选出LR(Logistic Regression,逻辑回归)模型。实验表明,优选出的基于LR模型的多特征融合疲劳检测算法,在检测的准确率、精确率以及召回率上都比现有的P80单特征检测,及基于模糊系统的检测算法有明显提高。
For information in English, please click here
Fatigue Detection Algorithm Based on Multi-visual Features Fusion by Using Machine Learning Model
Abstract:Fatigue detection is of great importance for reducing the rate of accidents. In this paper, a novel fatigue detection method based on machine learning is proposed to offset the shortness of existing fatigue detection methods. Eight visual fatigue features are calculated: blinking frequency, eyes closing period, yawning frequency, yawning period, nodding frequency, nodding period, turning frequency and turing period. Then the data set obtained was put into the machine learning models for training. LR (Logistic Regression) model shows better performance than other machine learning models. The lab result shows that the proposed multi-visual features based LR fatigue detection algorithm performs the best in recall, precision and accuracy compared with the existing fatigue detection methods.
Keywords: fatigue detection visual features machine learning LR model
基金:
论文图表:
引用
No.4674482112935514****
同行评议
勘误表
基于机器学习的多视觉特征融合疲劳检测
评论
全部评论0/1000