부산대학교 도서관

A fatigued driver is possibly considerably more dangerous on the road than a fast driver since he is prone to microsleeps. Researchers and manufacturers in the car industry are striving to solve this problem with a range of technical solutions aimed at preventing a catastrophe. A totally automated driver weariness state assessment system integrating driving video and a container-based approach is described, with a particular emphasis on tired driving detection In the suggested approach, the area of interest (ROI) is extracted using feature points, and face identification and feature point localization are accomplished using the convolutional network (CNN) architecture. to determine the condition of the eyes from ROI photos. The proposed CNN-based model might be used to develop a real-time driver drowsiness detection system for IoT devices that is both accurate and simple to use. This entire process is carried out within a container, which is a lightweight platform as a service that provides OS-level virtualization and portability. Surveillance has been elevated to a whole new level as a critical component of security measures with the introduction of signal processing technologies. The potential to give a lightweight alternative to heavier categorization models was achieved with Containerization and Dockerization. As a whole the model is to detect the drowsiness of the driver using IoT components, with Docker and Docker hub via cloud in a environment friendly way. The proposed model gives the prediction result with accuracy of 97% and the loss of 2.98%.