Resumen
To increase the reliability and efficiency of work in most information systems, biometric identification technologies are used, where the image of a person's face is used as an object of study. The aim of the study was to develop a biometric data validation model to provide fast and efficient verification and identification of users, as well as real-time control of their access. Computer vision technology and neural network technologies were used as research methods. To implement the biometric data validation model, we developed and studied our own ensemble of algorithms, designed on the basis of the BlazeFace basic model for face detection. In the face detection and recognition block, 478 facial points were identified, on the basis of which a three-dimensional face model was built. The found markers made it possible to calculate the parameters for validation: head tilt along different axes; gaze direction and eye openness; blurring of the analyzed frame. A distinctive feature of the proposed model is the emphasis on semantically significant areas of the face by performing additional calculations, which makes it possible to more accurately predict landmarks around the lips, eyes, and iris, increasing recognition accuracy. The results of a comparative analysis with the base model showed an increase in recognition accuracy by 14%, and a reduction in the total duration of face recognition by 3.6 times. The developed software can be used for video surveillance, monitoring the progress of tests in proctoring systems, as well as for ensuring security as part of the building security systems module.