Analysis of Sentiment Trends Based on Deep Learning

As the main expression medium of emotion, facial expression is used to explore personal emotion through facial expression recognition. The convolutional neural network architecture in deep learning can combine the emotion feature extraction and classification process, showing great advantages in emotion recognition by reducing the amount of computation and reference. In this paper, a deep learning-based facial expression recognition algorithm is proposed to realize expression analysis. The structure of multi-task convolutional neural network (MTCNN) is analyzed, the feature pyramid structure is introduced to design the C-MTCNN optimization algorithm, and the Inception v3 model is used to complete the feature extraction and emotion classification. The optimization of PCA algorithm and parameter adjustment on P-Net greatly improves the recognition accuracy of captured images, so as to realize the analysis of personal emotions. By testing the algorithm, the accuracy of the algorithm and the effect of the system are verified.