“The latest research report shows significant breakthroughs in the field of facial expression recognition. Researchers have effectively integrated spatial and temporal features by constructing a multimodal deep convolutional neural network, significantly improving the performance of facial expression recognition in videos. This method not only overcomes the shortcomings of traditional methods with low correlation between manual features and subjective emotions, but also demonstrates excellent performance in testing on public datasets, providing new ideas and methods for the development of video facial expression recognition technology.”