[论文解读] Driver Drowsiness Detection Model Using Convolutional Neural Networks Techniques for Android Application

Rateb Jabbar, Mohammed Shinoy+4 more|arXiv (Cornell University)|Jan 17, 2020

Sleep and Work-Related Fatigue参考文献 20被引用 195

一句话总结

本文提出一种基于轻量级CNN的驾驶疲劳检测模型，使用面部关键点在Android上实现，平均准确率超过83%，模型大小低于75 KB。

ABSTRACT

A sleepy driver is arguably much more dangerous on the road than the one who is speeding as he is a victim of microsleeps. Automotive researchers and manufacturers are trying to curb this problem with several technological solutions that will avert such a crisis. This article focuses on the detection of such micro sleep and drowsiness using neural network based methodologies. Our previous work in this field involved using machine learning with multi-layer perceptron to detect the same. In this paper, accuracy was increased by utilizing facial landmarks which are detected by the camera and that is passed to a Convolutional Neural Network (CNN) to classify drowsiness. The achievement with this work is the capability to provide a lightweight alternative to heavier classification models with more than 88% for the category without glasses, more than 85% for the category night without glasses. On average, more than 83% of accuracy was achieved in all categories. Moreover, as for model size, complexity and storage, there is a marked reduction in the new proposed model in comparison to the benchmark model where the maximum size is 75 KB. The proposed CNN based model can be used to build a real-time driver drowsiness detection system for embedded systems and Android devices with high accuracy and ease of use.

研究动机与目标

• 动机：需要实现实时、低资源的疲劳检测以防止微睡眠。
• 探索一种基于CNN的方法，利用面部关键点以提高对变化的鲁棒性。
• 开发一种适用于嵌入式/Android设备、无需繁重硬件的紧凑模型。
• 对比现有基准，评估准确性、模型大小和运行时性能。

提出的方法

• 对NTHU Driver Drowsiness数据集进行预处理，通过帧采样和数据增强提取600,000张图像。
• 使用Dlib/OpenCV在每张图像中检测68个面部关键点，以表示显著的面部区域。
• 提出 D2CNN-FLD：一个五层卷积神经网络，采用 Leaky ReLU 激活、丢弃（≈25%）、MaxPooling，以及用于两类分类的Softmax输出。
• 在归一化的关键点坐标上训练模型，使用最小-最大缩放和2类Softmax输出。
• 实现Android流水线，JNI将关键点数据从基于Java的应用传输到CNN进行实时推断。

实验结果

研究问题

• RQ1 基于面部关键点坐标的轻量级CNN模型是否能在NTHU数据集上达到具有竞争力的疲劳检测准确性？
• RQ2 提出的 D2CNN-FLD 在准确性和资源使用方面相对于先前的基于MLP和更重的CNN/RCNN基线表现如何？
• RQ3 照明条件和眼镜（眼镜/太阳镜）对检测性能有何影响？
• RQ4 该模型是否足够紧凑，能够在移动/嵌入式设备上实现实时运行且无需GPU加速？

主要发现

• D2CNN-FLD 在驾驶场景下实现了83.3%的整体准确率。
• 所提模型大小为0.075 MB（约75 KB），显著小于基准CNN/RCNN模型。
• 与先前的 D2MLP-FLD 相比，D2CNN-FLD 在汇总结果中精度提高约2.5个百分点。
• 检测时间随设备而异；例如，D2CNN-FLD 在NVIDIA Quadro P4000上为38.45 ms，在Samsung Galaxy S8 Plus上为142 ms。