A Comparative Analysis of Machine Learning Algorithms for Autonomous Face Mask Detection

Covid-19 pandemic is a global disease caused by severe acute respiratory syndrome. The rising number of infected individuals and death cases remain a major problem in 2021. Health protocols such as wearing a face mask is taken as prevention method to suppress the significantly growing numbers. Popular machine learning techniques have been addressed to assist in the global issue. This study intends to compare several popular classification algorithms namely K-Nearest Neighbors (K-Nearest Neighbors), Support Vector Machine (Support Vector Machine), Convolutional Neural Network (CNN), Decision Tree, and Naive Bayes for autonomous face mask detection as there are still limited sources of studies that does performance comparison of the related field. Experimental results are analyzed and evaluated using various measures such as precision, recall, accuracy, and F1 Score. Convolutional Neural Network proves to have the most promising performance than the other classification techniques to identify whether a person is wearing a mask or not with over 97% of accuracy. © 2022 Journal of Image and Graphics.

Design and Development Integrated Sensor System for Measuring Body Temperature and Heart Rate

Heart rate and body temperature are some of the important components of a person's main vital signs that need to be monitored regularly and periodically. The detection system technology continues to develop which allows a person to detect his own condition, to avoid exposure to COVID-19. However, the tools that are developing in the market are quite expensive and sometimes complex in operation because they are external products, so that it becomes a difficulty in itself. For this reason, it is important to design a detection device with sensor components that exist in the country and with a simple design so that it is easy to operate and inexpensive. In this paper, utilizing pulse sensors and AD8232 sensors to detect heart rate and MLX90614 sensors to measure body temperature, then NodeMCU ESP8266 to process sensor signals received and will be forwarded to the Display (LCD) to display the results carry out the design and development of an integrated sensor system. From the research results, the accuracy of the MLX90614 temperature sensor is very good with the achievement of 99.24% and the pulse sensor with the achievement of 98.86%. For the test results on each sample obtained accuracy values of 98.4% and 99% for the temperature sensor, and 92.3% and 92.2% for the pulse sensor, respectively. From these results, it is very clear that the sensor design deserves to be promoted as a quality product. © 2022 IEEE.