Face Recognition Program for Person Identification Based on Machine Learning Algorithms

In modern times, many companies use automatic systems for control access to internal premises to protect against unauthorized people. The most common solution in this task is to enter pass-cards that need to be attached to the readers in order to pass the turnstiles. Another way to organize an automatic pass control system is to create a personal identification system based on biometric data analysis. Today, the task of detecting objects, in particular faces, is already well studied, and the development of technologies allows us to train accurate models for solving face recognition task using GPU calculations instead of necessary using huge clusters. The purpose of this work is to study modern algorithms for detecting and recognizing faces in an image in order to develop a micro-service application for recognizing personality based on a face photo using machine learning algorithms. The application is implemented in the Kotlin 1.3 language and uses the OpenCV library as the backend for performing calculations in the neural network. Code for neural network training is written in Python 3.7 using the PyTorch 1.0 library for accelerating learning of neural networks on the GPU. This work contains 35 pages, 3 chapters, 10 figures, 3 tables, 5 formulas and 19 sources. Keywords: face recognition, machine learning, deep learning, neural networks, Siamese networks, face detection.