Corona Virus Disease is a new outbreak that can transmit infection through close contact or water droplets. Corona virus attacks the human respiratory system so that it can cause illness with symptoms of fever, cough and shortness of breath that can cause death. The use of a mask that covers the nose and mouth can prevent transmission. As a form of prevention, people are starting to be forced by regulations to always use masks in public places and when interacting with other people. However, it will be difficult for the authorities to monitor large groups of people. These problems can be solved with a system to detect masks. Mask detection in this study uses the naive bayes classification to distinguish a face with a mask correctly or incorrectly and also without a mask. The information used for classification is obtained through the histogram of facial image texture feature extraction using Local Ternary Pattern. The extracted image is preprocessed which includes resizing the image width and image grayscaling. The data used are 3,900 face images. Tests were carried out on the size of the image width, the threshold value, the number of bins, and the split of training and testing data. The results of the naive bayes classification produce an optimal accuracy of 68.462% with an image width of 50, a threshold value of 4, the number of bins 32, the distribution of training and testing data are 70%: 30%. Tests with 2 classes, namely correctly masked faces and unmasked faces, obtained an accuracy value of 86.15%. Based on these results, it is known that the naive bayes classification cannot properly classify images in the masked class incorrectly.
Copyrights © 2022