Research on psychological state assessment based on video sequences

HAN Haoxin; HUANG Wei; MAO Jiyong; LI Wangyang; HUANG Dong

doi:10.13276/j.issn.2097-1656.2024.11.005

Abstract：Objective To design and develop an algorithm and system for psychological state assessment research based on video sequences,so as to make up for the shortcomings of traditional evaluation process,provide an objective and fair evaluation process for evaluators,ensure that the evaluation method is not limited by time and region,and ultimately improve the efficiency of the evaluation process.Methods The evaluation system was mainly composed of the following three parts:facial image acquisition,psychological state assessment,and result visualization analysis.The facial image acquisition part was used to extract the video frames in the video and use the existing face segmentation algorithm to segment the face region in the image.In the psychological state assessment part,VGG was used to extract facial features,and the long short-term memory network was used to learn the temporal relationship between video frames to obtain psychological state results.The result visualization analysis part performed time backtracking of the final results in order to calculate the weight allocation of each frame image to the final results,and visualized the weight results.Results The database was divided into training set and test set at a ratio of 8:2,with the accuracy of 82.7％and the area under the curve of 0.89.Conclusion To a certain extent,individual psychological state assessment can be realized by the proposed method,which can solve the problem of individuals'increasing concern about mental health under the current social conditions,broaden the existing psychological state assessment approach,and provide a convenient,objective and efficient new assessment method for people who need psychological assessment.