A Laborer’s Mask-Wearing Behavior Detection Approach in the Manufacturing Field

Abstract

Industry 4.0 has considerably advanced multiple manufacturing fields through digitalization and intelligentization. Many technologies, such as supervisory control, data acquisition, and data analytics, have been used widely in manufacturing sites to enhance production efficiency. Therefore, this created a cloud-based anomaly detection module for epidemic prevention at the manufacturing site. Image process technologies, deep learning algorithms, and cloud computing were employed in the proposed module to automatically identify labor anomaly behavior in the manufacturing site and prevent the epidemic. This study used image processing technologies and deep learning to recognize and train the manufacturing site image. Accordingly, the analyzed result could be incorporated into the cloud system using the Message Queuing Telemetry Transport (MQTT) protocol. Therefore, the administrators and laborers can be notified regarding the anomaly behavior. The author used the image data obtained from the cylinder head process site as a data source for DA. As per the experimental results, the proposed method has an accuracy of 90%. Therefore, deep learning algorithms provide a practical approach to anomaly detection for epidemic prevention. Furthermore, this study’s primary contributions are designing an improved approach and connecting the manufacturing site to the cloud side using the proposed module.