Machine Learning Techniques for Real-Time Human Face Recognition

The development of real-time facial recognition software continues to surge forward. Uniquely recognizing human faces in a real-time system relies heavily on face detection and recognition. When it comes to authentication and other forms of security, the face is where it’s at. An improved and faster facial detection system is a primary goal. This work introduces a CNN and Python-based Face Recognition System. The paper presents the analysis of machine learning classification techniques to identify leading predictive algorithms. Further, the algorithms namely Decision Tree, Naïve Bayes, KNN, and CNN analyzed. In the proposed work OpenCV and Python are applied to the dataset after pre-processing images. For this purpose, a celebrity dataset of faces is utilized. In addition, a face or faces caught in the live feed are identified. The process considered two phases for the face recognition system: the training phase and the testing phase. Eighty percent of the human face samples are learned during training, and twenty percent of the data is used for testing. An accuracy of 89.36% is achieved by using machine learning to improve accuracy measures including recall value, f-score, and precision. Comparative performance analysis of these machine learning techniques also performed.