Facial expression recognition method based on spatiotemporal attention mechanism in museum scenes
Museums are important places for the public to learn about local culture and history.With the development of digi-tal technology and artificial intelligence,visitors'browsing of museums is no longer a one-way process,but rather an interaction between visitors and exhibits,The first step in achieving bidirectional interaction is to understand the expressions of tourists,in order to make correct judgments and responses to their current state.Research was conducted on tourist expression recognition in museum scenes and a facial expression recognition method based on spatiotemporal attention mechanism was designed,which extracts facial expression features from both temporal and spatial dimensions using convolutional neural networks(CNN)and long short term memory networks(LSTM).Meanwhile,adding spatiotemporal attention mechanism enhances feature expression ability.Based on the experimental results,the highest accurate recognition rate can reach 78%for six different facial emotions of visitors.The experimental results on the publicly available RML(ryerson multimedia lab)video sentiment dataset show that recognition method achieves a correct recognition rate of 64.63%.The experimental results show that recognition method can effectively im-prove the expression recognition rate.
convolutional neural networkslong short term memory networksfacial expressionspatiotemporal attention
万方数据
维普
