Application of deep learning and machine learning models to detect COVID-19 face masks - A review

Abstract

The continuous COVID-19 upsurge and emerging variants present unprecedented challenges in many health systems. Many regulatory authorities have instituted the mandatory use of face masks especially in public places where massive contact of people is frequent and inevitable, particularly inside public transport facilities, sports arenas, shopping malls and workplaces. However, compliance and adherence to proper wearing of face masks have been difficult due to various reasons including diversified mask types, different degrees of obstructions, various variations, balancing various model detection accuracy or errors and deployment requirements, angle of view, deployment of detection model on computers with limited processing power, low-resolution images, facial expression, and lack of real-world dataset. Therefore, this study aimed at providing a comprehensive review of artificial intelligence models that have been used to detect face masks. The study revealed that deep learning models such as the Inceptionv3 convolutional neural network achieved 99.9% accuracy in detecting COVID-19 face masks. We deducted that most of the datasets used to detect face masks are created artificially, do not represent the real-world environments which ultimately affect the precision accuracy of the model when deployed in the real world. Hence there is a need for sharing real-world COVID-19 face mask images for modelling deep learning techniques. The study also revealed that deeper and wider deep learning architectures with increased training parameters, such as inception-v4, Mask R-CNN, Faster R-CNN, YOLOv3, Xception, and DenseNet are not yet implemented to detect face masks.

Control and Prevention of SARS-CoV-2 Outbreaks among Healthcare Workers from 129 Healthcare Facilities in Mexico

“…Different artificial intelligence-related approaches have been tested to detect face mask use and body temperature measuring [39][40][41]. Some difficulties remain in that field, such as the detection of different mask types, different degrees of obstructions, among others [42]. However, that technology could help rapidly improve the contact-less detection of suspicious COVID-19 cases and similar respiratory diseases, such as flu.…”

Are we there yet? Unbundling the potential adoption and integration of telemedicine to improve virtual healthcare services in African health systems

“…The virus was declared a public health emergency and a global public health concern by the World Health Organization (WHO) on January 30th, 2020 [ 1 ]. Immediately after the pronouncement, many countries imposed stringent measures including social distancing, the mandatory wearing of face masks in public places [ 2 ], self-isolation and quarantine of suspected and positive individuals, a temporary ban of public gatherings [ 3 ], closure of borders, sanitization, temperature checking strategic entry points [ 4 ] and curfews to reduce further transmission and the catastrophic impact of the virus. However, COVID-19 measures like lockdowns, self-isolation, quarantine and stay-at-home with unintended consequences, such as widening economic inequalities, abuse of substances [ 5 ], mental health problems [ 6 ], and exacerbating poor medical outcomes.…”

Hybrid Deep Feature Generation for Appropriate Face Mask Use Detection

“…Such systems can also serve as adjunct tools to be used by clinicians to confirm their findings. Machine learning methods have been used to detect face masks automatically [ 32 , 33 ]. A wide range of deep learning models, especially convolutional neural networks (CNNs) [ 34 , 35 ], were used to solve computer vision problems.…”