Application of machine learning in Artificial Intelligence
The covid-19 epidemic, which has become a significant problem not only in Sri Lanka but also around the world, has already claimed the lives of millions of people. There are various vaccines and drugs for this epidemic disease, but no one is responsible for the future side effects of these solutions. Therefore, ‘prevention is better than cure’ will be the most suitable option in this current situation. The World Health Organization has announced that wearing appropriate face masks according to the health instructions can control the disease. The relevant authorities are under significant pressure, and they are working day and night to curb this menace to promote wearing masks in the correct manner. Although the appropriate authorities are trying to explain this to the citizens, some people do not care about the instructions and are endangering themselves and others. The number of deaths is increasing day by day, and some countries have to lockdown because of this situation, and this will severely affect the entire world economy. Many countries have enacted strict laws to wear face masks to reduce the damage in this current situation. Developing countries such as Sri Lanka have a severe impact on the economy because of lockdown. Therefore, it is essential to follow the relevant health advice, such as wearing masks to face the current situation. It is necessary to trap people who do not wear face masks according to safety standards and act against them. This system is beneficial for the health authorities, security forces in public areas and health organizations to facilitate those in the field.
Correctly wearing a mask can be introduced as a solution to control the spread of Covid-19 contagious disease. If one infected person is socialized without wearing the mask, the infection spreads rapidly. However, monitoring and detecting anyone who is not wearing a face mask on a large number of groups at once becomes a difficult and a big task. Also, some people wear masks, but it is a significant problem for the relevant authorities to identify a large number of groups in one instance whether they are wearing masks according to the prescribed health instructions or not. This real-time face mask monitoring system can be used in existing CCTV cameras with computer vision techniques to monitor people and solve the current epidemic.
The main aims of this system are to reduce the number of deaths due to the covid 19 epidemic in the world today, Capture the people who are not wearing masks or those who are not wearing incorrect procedure and reduce the risk of human deaths, Support to the relevant authorities to trap people who break the instructions and to normalize the day to day affairs of the world while the epidemic continues.
The methodology of the system can be described under several topics. Data processing and training of the model are two tasks of real-time face monitoring. Under data processing it can be done using data visualization, conversion of RGB image to gray image, and finally reshaping the data. After processing and splitting the data, the model has to be trained. When considering face detection, first need to apply face detection to compute the bounding box and then have to identify the facial landmarks and detect the eyes, mouth, etc. This process is very important to identify the persons’ face mask validation. Next, by getting deep learning knowledge, we have to create the dataset of wearing masks and train the machine to identify the faces with masks and without masks. Training (i.e., educating) the face mask detector will be followed in several steps. First, have to train a face mask detection with mass images and after training the system has to again retrain the machine with the rejected images giving the set of conditions. After that have to do image differentiation to detect mainly two categories (with mask and without mask). If someone is wearing masks, then have to follow the conditions and check whether the mask is according to the instructions. If a person is without masks or not following the health instructions then the alert system process will take place. The system will capture a photo of the person who is not according to the safety standards and add the image to the interface and then generate a report using the real-time monitoring dashboard. Next, using the notification interface, the system will send the details of the person with the location as an AI-based alert message to the admin.
The originality of this system is some research worked on the projects to identify only the face mask, but that will fail according to the current situation. To prevent the disease people should adhere to the relevant health instructions. Therefore detecting only a face mask is not enough. Some other researchers have proposed systems to identify whether masks are in the correct position. However, in those systems, researchers have only worked on implementing the invention that allows individuals to check whether they are wearing their face masks according to the correct instructions or not on their own. However, they have not implemented this system to cover the public areas to identify the validity of the mask among a group of people. There are indeed systems to identify face masks using CCTV cameras, web cameras, etc., to identify the face masks of a group of people. But there is no system yet to identify the validity of the face masks among public places. Many researchers explored CCTV systems with SMS alerts. Therefore, those system features can be used in the proposed system to combine and trap people who break the health instructions. Therefore, this system offers to tackle all these systems and develop an effective system to solve the current global problem.
0 Comments