Face Mask Detection using Machine Learning Techniques

Reddy, P. Dwarakanadha; Nandini, M; Mamatha, E.; Reddy, K. Vijaya Kumar; Vishant, Anand

doi:10.1109/icoei51242.2021.9452826

Cited by 3 publications

(1 citation statement)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, it is aimed to detect unmasked persons over a video using this developed system. From the results, it is seen that people wearing and not wearing masks can be successfully detected [4]. A system has been proposed to detect persons not wearing masks.…”

Section: Introductionmentioning

confidence: 96%

Face Mask Detection Using GoogLeNet CNN-Based SVM Classifiers

Sünnetci

Akben

KARA

et al. 2023

Gazi University Journal of Science

View full text Add to dashboard Cite

The COVID-19 pandemic that broke out in 2019 has affected the whole world, and in late 2021 the number of cases is still increasing rapidly. In addition, due to this pandemic, all people must follow the mask and cleaning rules. Herein, it is now mandatory to wear a mask in places where millions of people working in many workplaces work. For this reason, artificial intelligence-based systems that can detect face masks are becoming very popular today. In this study, a system that can automatically detect whether people are masked or not is proposed. In the dataset used in the study, there are a total of 440 images, 220 with masks and 220 without masks. Here, we extract image features from each image using the GoogLeNet deep learning architecture. Using the loss3-classifier layer of the GoogLeNet deep learning architecture, we obtain 1000 deep image features for each image. With the help of these image features, we train GoogLeNet based Linear Support Vector Machine (SVM), Quadratic SVM, and Coarse Gaussian SVM classifiers. Afterward, we develop a Graphical User Interface (GUI) application of the proposed system and a system that can detect masks from snapshots by saving the trained models. Thus, an application that is based on artificial intelligence and can automatically detect the mask has been developed. From the results, it appears that the accuracy (%), sensitivity (%), specificity (%) precision (%), F1 score (%), and Matthews Correlation Coefficient (MCC) values of GoogLeNet based Linear SVM, Quadratic SVM, and Coarse Gaussian SVM classifiers are equal to 99.55-99.55-99.55-99.55-99.55-0.9909, 99.55-99.55-99.55-99.55-99.55-0.9909, and 99.55-99.09-100-100-99.54-0.9910, respectively. When the results are examined, it is seen that not all classifiers are superior to each other in terms of accuracy. However, while Linear SVM and Quadratic SVM classifiers are more successful in terms of sensitivity and F1 score, it is seen that the Coarse Gaussian SVM classifier is better in terms of specificity, precision, and MCC.

show abstract

Section: Introductionmentioning

confidence: 96%