<p>Video data are especially useful for facial recognition because they capture detailed and constantly changing facial features, making it possible to improve recognition accuracy. To recognize facial expressions in video data, this study proposes a facial expression recognition model with feature-enhanced convolutional networks. The model first uses a multi-task convolutional neural network for face alignment and then utilizes a feature-enhanced convolutional network for facial expression recognition. When the dataset size was around 500, the accuracy of the multi-task convolutional neural network model was 0.95, and the feature point error value was 0.02. As the iteration reached 200, the recognition accuracy of the facial expression recognition model based on a feature-enhanced convolutional network was 0.97. When the validation set size was 1000, the accuracy of the facial expression recognition model with the feature-enhanced convolutional network was 0.84. Therefore, the proposed model can provide a novel and efficient technical route for the field of facial recognition and expression recognition.</p>