A new strain of virus was identified in humans, known as novel coronavirus (nCoV), which was never previously been identified in humans. Coronaviruses (CoV) are a wide group of viruses which cause illness that range from basic colds to infections like Middle East Respiratory Syndrome (MERS) and Severe Acute Respiratory Syndrome (SARS). The first infected patient of coronavirus was found in December 2019. The habit of wearing face masks while stepping out is rising due to the COVID- 19 corona virus epidemic. Before Covid-19, masks were worn by people to protect their health from air pollution. Scientists have concluded that wearing face masks works on decreasing COVID-19 transmission. In 2020, the rapid spread of COVID-19 led the World Health Organization to declare COVID- 19 as a global pandemic. The virus spreads through close contact of humans and in crowded/overcrowded places. Among them cleaning hands, maintaining a safe distance, wearing a mask, refraining from touching eyes, nose, and mouth are the main, where wearing a mask is the simplest one. Unfortunately, people are not following these rules properly which is resulting in speeding the spread of this virus. The solution can be to detect the people not wearing mask and informing their authorities. the face mask detection is a technique to find out whether the person is wearing a mask or not. In medical applications Deep learning techniques are highly used as it
allows researchers to study and evaluate large quantities of data. Deep learning models have shown great role in object detection. These models and architectures can be used in detecting the mask on a face. Here we introduce a face mask detection model which is based on computer vision and deep learning. The proposed model can be integrated with computer or laptop cameras allowing it to detect people who are wearing masks and not wearing masks. The model has been put together using deep learning and classical machine learning techniques with opencv, tensor flow and keras. We have introduced a comparison between three machine learning algorithms to find the most suitable algorithm that yields the highest accuracy.
The spread of COVID-19 virus has reduced but it is still not over. If everyone follows all the safety measures, then it can come to an end. This will help in lowering the cases to such a level that COVID19 virus can vanish from everywhere.
In 2020, the largest pandemic in recent history spread through the world: COVID-19. As of May 1st, 2021, there have already been 152 million cases and 3 million deaths around the world [1]. In many regions, those numbers are considerably under-counted [8]. Beyond that, many parts of the world have slowed or stopped due to the human, economic, and social impacts of distancing and protection measures. For the purpose of the ongoing pandemic and predictions for future pandemics [13], this project seeks to create a mask detection system that is capable of recogniz- ing whether people in surveillance-type video streams are correctly wearing their masks. Face-mask detection represents both a detection as well as a classification problem because it requires first the location of faces of people in digital images and then the decision of whether they are wearing a mask or not. The first part of this problem has been studied extensively in the computer vision literature, due to the broad applicability of face-detection technology [6]. The second part, on the other hand (i.e., predicting whether a face is masked or not), has only gained interest recently, in the context of the COVID-19 pandemic. Although a considerable amount of work has been done over the last year on this part [7–17], it typically only tries to detect whether a mask is present in the image. No special attention is given to whether the masks are properly placed on the face and are, hence, worn in accordance with the recommendations of medical experts. This limits the application value of existing face-mask detection techniques and warrants research into computer vision models capable of not only detecting the presence of facial mask in images, but also of determining if the masks are worn correctly.
To further illustrate this issue, several sample images from the MAFA (MAsked FAces) dataset [18] are shown in Figure1. MAFA represents one of the most popular datasets for training face-mask detectors, and in a similar manner to most other datasets publicly available for this purpose, contains only binary labels indicating whether face-masks are present in the images or not. As illustrated in Figure1masked facial images in such datasets typically belong to one of two groups: (i) faces with correctly worn masks (marked green), and (ii) faces with incorrectly worn masks (marked red). Because the correctness (or compliance with recommendations) of the mask placement is not annotated, existing mask detectors commonly learn from highly noisy data (considering the intended purpose of the detectors) and commonly do not flag faces where a mask is present, but does not cover the nose, mouth and chin. We note that this is a common issue seen across most of the existing work on face-mask detection [16,19–21] and has important implications for the usefulness of the designed detectors in practice