Prototype Model Development using Python for Detection of Transparent Face Mask and Identification of the State of Usage of Transparent Face Mask

Abstract

Purpose – The right use of face masks is one of the health practices that have been implemented in response to the COVID-19 pandemic danger. The automated monitoring of face mask usage is aided using face mask detection models. 

Methodology – A prototype model that can detect and identify the state of usage of a transparent face mask was developed using the object detection model Resnet50. It was trained with datasets containing people wearing proper and improper way of transparent masks and people who do not wear masks. The proponents also employed OpenCV to enable computer vision, which is utilized for methods of image processing like reading and resizing.

Result – The prototype model's overall accuracy and precision were tested on 19 distinct test scenarios, with the prototype model obtaining an average of 50.21% accuracy and 49.96% precision. The worst-case scenario for the prototype model's responsiveness is 2 frames per second.

Conclusion – The prototype model was developed as part of a pilot study to detect and identify the state of use of a transparent face mask. The test demonstrated that the prototype model can detect and identify the state of use of transparent face masks, namely proper, improper, and no mask, accurately, precisely, and responsively.

Recommendation – Future researchers must consider utilizing new techniques to further train the model; this may be accomplished by adding more images to the dataset as well as applying various types of pre-trained/hybrid models that are more optimal.

Practical Implications – The prototype model promotes health security and public safety and adheres to standard data privacy protections in the usage of a transparent face mask.