Emotion Modeling and User Experience Enhancement of Digital Media System in VR Environment

This study addresses emotion modeling and user experience enhancement within digital media systems in a virtual reality (VR) environment. A novel spatio-temporal neural network architecture that integrates asymmetric non-local elements and an efficient channel attention mechanism is proposed. This architecture improves emotion recognition accuracy by fusing high- and low-order information. In addition, a spatio-temporal LSTM module is introduced to capture spatial and temporal correlations in feature maps and image sequences, respectively. Experimental results demonstrate the effectiveness of the proposed approach, achieving an average accuracy of 97.68% and an average F1 score of 0.653 in sentiment analysis tasks. Thus, the model provides an end-to-end solution for emotion recognition and contributes to the advancement of digital media systems in VR environments.