The Application of Computer Technology in College Students' Mental Health Education and Psychological State Recognition and Early Warning

Abstract

To address the challenge of traditional questionnaires failing to accurately reflect individuals' psychological states, this paper proposes the integration of computer technology into mental health education for college students to identify psychological states and issue timely warnings. The study adopts an improved pyramid optical flow algorithm combined with a micro-expression recognition algorithm based on three-dimensional convolutional neural networks (3D-CNN). This combination further enhances the technical robustness. First, preprocessing algorithms are used to crop facial images; then, the pyramid optical flow algorithm is employed to extract facial features from the images, capturing dynamic characteristics of micro-expressions. Based on these steps, 3D-CNN is used to train the feature data, achieving efficient recognition of micro-expressions. Experimental results show that the micro-expression recognition accuracy on the CASME dataset reached 89.1%, with an F1 score of 0.6742, outperforming other traditional methods. The proposed algorithm significantly reduces model training parameters and computation time while exhibiting stronger feature learning capabilities. By accurately identifying students' psychological states, it provides objective data support for mental health education in colleges and, more importantly, offers an effective solution for early warning of psychological problems.