SOFTWARE SUITE FOR CLASSIFYING HUMAN EMOTIONS BASED ON FACIAL IMAGES

This article is devoted to the development of a software suite for the automatic classification of human emotions based on facial images, analyzing modern computer vision and machine learning technologies. Emotion classification plays a crucial role in improving human-computer interaction, developing intelligent agents, and solving automation-related tasks in fields such as medicine, psychology, security, education, marketing, and robotics. Within the scope of the research, algorithms used for analyzing emotional states through facial expressions were examined, particularly convolutional neural networks (CNNs), models based on facial action units, spatiotemporal dynamic analysis methods, and multimodal approaches. Algorithms aimed at increasing accuracy through various visual contexts and the movement of facial landmarks, based on P. Ekman's universal emotional states, were also studied. The primary goal of the study is to develop an intelligent software suite capable of classifying emotional states based on data extracted from facial images, to select the most effective approach based on an analysis of existing classification models, and to evaluate the possibilities of its practical application. In this regard, special attention is paid to analyzing the dynamics of visual features on the human face, enriching them with contextual information, and selecting optimal architectures to achieve high accuracy. By providing an in-depth analysis of existing scientific approaches in the field of emotion classification, the research serves to identify their advantages and shortcomings, thereby creating a scientific foundation for future studies.

Hali tarjima qilinmagan