The study is devoted to an important direction of AI - the application of deep learning of neural networks for face recognition. The text focuses on the effectiveness of siamese neural networks in face recognition. Attention is paid to the improvement of one of the most popular varieties of Siamese models - FaceNet model. The technology for creating fine-tuning of neural network models for training on given data sets has been applied. To solve the problem of recognizing people from their images, the training of the Siamese model was carried out using the proposed settings and their effectiveness was evaluated. A face recognition system was implemented using a fine-tuned model to demonstrate the effectiveness of the proposed method.