AV-ITN: A Method of Multimodal Video Emotional Content Analysis

With the rapid development of Internet technology, social media has also developed rapidly with the support of the Internet. Social media data has grown exponentially. A large amount of data needs to be reviewed urgently. The method of machine review requires video emotional content analysis technology (Affective Video Content Analysis Technology). Analysis, AVCA) support. A large number of studies have shown that the use of deep learning methods to achieve emotional content analysis is currently the most effective method, and its effect is better than traditional algorithms and manual methods. Based on this, this paper proposes a multi-modal video sentiment analysis algorithm AVITN that utilizes both audio and video modal information to promote sentiment analysis in videos. AV-ITN achieves a high accuracy rate of 83.66% on the IEMOCAP benchmark. The multimodal video sentiment analysis algorithm proposed in this paper can obtain more emotional information contained in the video and effectively improve the accuracy of video sentiment analysis.