Usually, the cognition of patients with depression will have several obvious characteristics. With the development of medicine, the cognitive analysis of depression gradually changes to the analysis by extracting the characteristics of patients' emotion, expression, voice, brain, behavior and so on. For example, Vidal Ribas P et al. Concluded that there was a significant horizontal and vertical relationship between emotion recognition disorder and youth severe irritable depression symptoms through experimental control analysis [16]. Pan Wei et al. 2018 used machine learning method to recognize depression by analyzing the relationship between depression and speech. The experimental results show that the recognition rate of depression reaches 82.9% [17]. Yang Hong et al. Identified depression by expression and behavior in 2018 [18]. Ma Lin and others proposed a method to judge the degree of depression based on the expression by analyzing the expression differences between patients with depression and normal people. The results showed that the recognition rate of expression in patients with depression was lower than that in normal people [19]. However, because most people do not have a deep understanding of psychological problems, many people do not realize that they are suffering from depression, and some people are biased against depression and other mental diseases and are not willing to seek treatment. As a result, few patients take the initiative to diagnose, so the corresponding information and features are not easy to extract. At the same time, depression is a relatively sensitive mental disease, and this face-to-face information acquisition may cause further harm to patients.