Abstract
Our first symptoms in COVID-19 patients are fever, sore throat and runny nose. These symptoms have become indicators for the public to determine who may be infected.
The health assessment system uses facial images as input and predicts whether the person in the image is healthy, has a fever, a sore throat, or is bleeding. The project was developed using traditional machine learning techniques. Four video extraction methods combined with four machine learning classes were tested and analyzed to find the best models to integrate into the predictive health UI. Face extraction techniques used are local binary model (LBP), principal component analysis (PCA), linear regression (LDA) and Gabor filter. The classes used are support vector machine (SVM), neural network (NN), k-nearest neighbors (KNN), and random forest (RF). The best overall model selected for the UI of health prediction is the LBP + NN model with the highest average score of 76.84% for secondary distribution. It also performs very well in single-level classification, fitting less than other similar models, and achieving average training and testing accuracy of 94.38% and 86.87%, respectively. The main goal of Physiognomy Precision is to create a new concept that uses the face to focus on health. The project aims to create a seamless, efficient and effective system for the early detection and care of many conditions by extracting valuable health-related information from facial images. This project uses machine learning algorithms to train models on different datasets of facial images and related medical data. Specifically, these models learn complex patterns and relationships between faces and health, ensuring they are as accurate as new faces. It is very important to ensure the use of facial information and health prediction. The project strictly adheres to the principles of agreement, transparency and user control. Users will be given clear information on how their facial information will be used, and stringent measures will be put in place to prevent misuse or abuse.
The integration of facial recognition technology with traditional machine learning techniques represents a pioneering approach in the realm of health assessment. By leveraging methods such as LBP, PCA, LDA, and Gabor filter, alongside SVM, NN, KNN, and RF classifiers, the project achieves remarkable accuracy in predicting health indicators from facial images. This innovative initiative, known as Physiognomy Precision, not only aims to revolutionize early detection and treatment but also underscores a commitment to ethical practices. Through transparent communication and robust safeguards against misuse, the project ensures user trust and privacy, paving the way for a future where facial information enhances personalized healthcare with integrity.