基于深度置信网络和生成模型的驾驶疲劳识别

Driver Fatigue Recognition Based on Deep Belief Network and Generative Model

驾驶疲劳识别研究对预防交通事故提高交通安全具有重要意义.提出了一种基于深度置信网络和生成模型的驾驶疲劳识别方法.为了有效地表示疲劳,采用深度置信网络从人脸图像中提取疲劳特征;结合已标注样本和未标注样本,提出了一种基于生成模型的半监督学习的疲劳识别方法,解决了疲劳识别中的小样本问题.在自建疲劳数据库上,采用该方法进行了驾驶疲劳识别的仿真实验,同时和其他几种方法进行了对比,结果表明该方法具有更高的识别精度.

Driver fatigue recognition has great theoretical significance and applied value in reducing accidents and improving traffic safety.A novel method based on deep belief network and generative model is proposed for driver fatigue recognition.In order to represent fatigue effectively,fatigue features are extracted using deep belief network（DBN）from the facial image.The semi-supervised learning method for fatigue recognition based on generative model is proposed to solve the problem of small sample in recognition,which makes use of both the labeled and unlabeled samples.Experiments were performed on self-built database to test and evaluate the proposed method.The experiment results show that our method has higher recognition accuracy than other state-of-the-art methods.