Visual media makes image analysis more important and complicated. Images and photos are increasingly used to provide information to the web's enormous audience. Start using photos to understand image propagation patterns. Images, like words, should evoke complicated emotions. Unlike text, where semantics and context are easily accessible for sentiment analysis, extracting and understanding image sentiment is difficult. This paper presents the intermediate characteristics-based picture emotional tone prediction algorithm. Thus, feature selection methods yield simpler results than using the image's low-level attributes. To improve face-containing picture accuracy, it includes many facial expression detection algorithms. The exploratory investigation shows that the proposed framework outperforms earlier techniques in predicting accuracy. Analyzing the predictions could also reveal the relationships between intermediate characteristics and visual emotion.