【摘要】 针对驾驶员疲劳驾驶检测困难及检测精度低等问题，提出基于改进YOLOv5s的疲劳驾驶检测方案，以提高终端的边缘智能识别能力。以YOLOv5s为基础框架，通过改进损失函数，提高模型精度与鲁棒性；通过添加注意力机制模块，提高算法的特征提取能力和检测精度。开展基于改进EIoU损失函数和添加CBAM注意力模块的消融试验。试验结果表明：基于改进YOLOv5s的疲劳驾驶检测准确率和召回率分别为95.2%和95.0%,相较于原始YOLOv5s模型，闭眼检测精度提高了3.6%,哈欠检测精度提高了3.8%。更多还原

【Abstract】 Aiming at problems such as the difficulty of driver fatigue driving detection and low detection accuracy, the fatigue driving detection scheme with improved YOLOv5s is proposed to improve the edge intelligent recognition capability of the terminal.Taking YOLOv5s as the basic framework, the model precision and robustness are improved by improving the loss function; The attention mechanism module is added to improve the feature extraction ability and detection precision of the algorithm; The ablation test of improving EIoU loss function and adding CBAM attention module is carried out, and the test shows that the accuracy and recall of the improved YOLOv5s in fatigue driving detection are respectively 95.2% and 95%,and compared with the original YOLOv5s model, the eye closure detection precision is improved by 3.6%,and the mouth yawn detection precision is improved by 3.8%.更多还原