Drowsiness on the road is a widespread problem with fatal consequences; thus, a multitude of solutions implementing machine learning techniques have been proposed by researchers. Among existing methods, Ghoddoosian et al.'s drowsiness detection method utilizes temporal blinking patterns to detect early signs of drowsiness. Although the method reported promising results, Ghoddoosian et al.'s algorithm was developed and tested only on a powerful desktop computer, which is not practical to apply in a moving vehicle setting. In this paper, we propose an embedded system that can process Ghoddoosian's drowsiness detection algorithm on a small minicomputer and interact with the user by phone; combined, the devices are powerful enough to run a web server and our drowsiness detection server. We used the AioRTC protocol on GitHub to conduct real-time transmission of video frames from the client to the server and evaluated the communication speed and processing times of the program on various platforms. Based on our results, we found that a Mini PC was most suitable for our proposed system. Furthermore, we proposed an algorithm that considers the importance of sensitivity over specificity, specifically regarding drowsiness detection algorithms. Our algorithm optimizes the threshold to adjust the false positive and false negative rates of the drowsiness detection models. We anticipate our proposed platform can help many researchers to advance their research on drowsiness detection solutions in embedded system settings.
翻译:路边的沉睡是一个具有致命后果的广泛问题;因此,研究人员提出了实施机器学习技术的多种解决方案;在现有的方法中,Ghoddoosian等人的沉睡探测方法使用了时间闪烁模式,以探测潜伏的早期迹象。虽然这种方法报告了有希望的结果,但Ghoddoosian等人的算法仅是在强大的台式计算机上开发并测试的,该算法不适用于移动车辆设置。在本文中,我们提出一个嵌入系统,能够在小型微型计算机上处理Ghoddoosian的沉睡探测算法,并通过电话与用户进行互动;加在一起,这些装置足够强大,足以运行一个网络服务器和我们的沉睡探测服务器。我们使用GitHub的AioRTC协议实时将视频框架从客户传送到服务器,并评价各个平台上程序的通信速度和处理时间。根据我们的结果,我们发现一个微型PC最适合我们提议的系统的潜伏探测测算方法;此外,我们提议了一个具有超常性测算能力的系统测算模型,以反向性测算法的精确度测算速度速度,具体地计算。我们提出的测算的测算模型的测算方法在精确测算中,我们测算模型上了我们的测算的精确测算方法的精确测算方法的精确测算。