MODEL FOR PERSON PSYCHOEMOTIONAL STATE DETERMINATION USING AUDIO AND VIDEO DATA

Determination of the person’s psychoemotional state finds application in a variety tasks from diagnosis diseases to prevention emergency traffic situations. Given importance and where it used, there are many methods approaches exists, but most them use only one nonverbal feature, such as person's voice or facial expressions. At same time, simultaneous several signs can increase accuracy compared using sign. In this paper, method has been developed that uses expression signs. Based on literature review analysis existing methods, best were selected, was with solutions. The made possible improve CMCNN, at each test, Accuracy its own less comparison DisVoice + SVM. For improving weighted average proposed instead usual more complex model, which is able detect interrelation between expression.