Advanced Customer Emotion Detection in Online Shopping: A CNN and BiLSTM-Based Approach

Creating a mimicked environment while shopping online is becoming more common, and the majority of transactions are now conducted online. Recent innovations in e-commerce and business have made it critical to offer personalized online suggestions based on emotion prediction to users or consumers. A person's facial emotions are easy to read, but creating an algorithm that can accomplish the same thing is a huge challenge. Using facial expressions as a basis for emotion recognition, this proposed suggests a new approach for online recommendation systems. The proposed approach includes three steps: detecting faces, extracting facial features, and training the model. Because every person's face is unique, using facial detection to just identify when someone is in the room can be a challenging undertaking. In order to change the look of a person's face, whether permanent or temporary, facial feature extraction is performed. When the facial muscles are contracted, it affects the direction and amplitude of skin surface displacement. The proposed approach used CNN-BiLSTM for the model's training. This cutting-edge method outperforms BiLSTM and CNN with an average accuracy of 91.62%.