This project intends to study a multimodal emotion recognition method. This project aims to provide a theoretical basis for the problem of emotional feedback in the process of mental health education, which will help teachers and counselors to monitor students’ emotional status in real time and provide effective guidance. Multimodal information such as facial expressions, voice and text can be used to accurately identify emotions and accurately feedback on mental health. This project intends to use multi-source information fusion technology to extract and model emotion features and improve the accuracy and robustness of emotion recognition. The fusion of multi-level feature information improves the ability to recognize complex emotions. Simulation experiments show that the accuracy of the algorithm in emotion recognition is 89.7%, which is about 12.5% higher than the traditional emotion recognition algorithm based on a single mode. In addition, this method is balanced in the classification of emotion types, especially in the recognition of negative emotions, and can be used as a means of emotion monitoring and intervention.
Analysis of Emotion Recognition Model Based on Multimodal Deep Neural Network Algorithm
2024-10-23
575592 byte
Conference paper
Electronic Resource
English
Deep spatio-temporal feature fusion with compact bilinear pooling for multimodal emotion recognition
British Library Online Contents | 2018
|Multimodal Emotion Recognition and Intention Understanding in Human-Robot Interaction
Springer Verlag | 2021
|