The use of computer vision to determine the geometry of a face using reference points is a relatively new approach in medicine. The relevance of this study is due not only to the need to develop new methods and approaches in determining the geometry of the face, but also to the growing interest in the development and application of artificial intelligence in medicine. The research objective. The purpose of this article is to develop mathematical and neural network algorithms that determine the geometry of a face using reference points. Material and methods. To train the neural network, a small sample of 1,000 marked-up photos in the public domain was used, which depict a full-face portrait of a man up to his shoulders. Only adult (18+) representatives of the Caucasian race were considered. The photos were marked up using the LabelImg 1.8.6 graphical image analysis tool, in which the areas of finding (classes of definition) of reference points were manually marked in the graphical interface mode. YOLO 8 was chosen as the neural network architecture. Results. It was shown that the accuracy of the point search by a neural network trained on 1000 grouped photographs reaches 98.15 %, which indicates a good definition of the boundaries of objects and their classification. Conclusion. The conducted research demonstrated the successful development and application of neural network algorithms to determine alternative reference points characterizing the geometry of the face. The results obtained confirm the effectiveness of the technique proposed by the authors in cosmetology, and also indicate its potential for use in other fields of medicine, such as dentistry, neurology and surgery.