Real-time Facial Emotion Recognition using CNN

Abstract

Visual sentiment analysis explores how people emotionally respond to visual content like images and videos. It aims to decipher the underlying emotional context by examining visual data. The success of this field largely stems from advances in computer vision technology, leading to robust algorithms that extract meaningful information from visuals. Many models in this field focus on whole-image features or develop intricate architectures to analyze visual sentiment. However, these approaches often overlook the significance of local areas within images, which can be crucial in determining emotional responses. This research paper addresses this gap by applying Convolutional Neural Networks (CNNs) to detect human faces in images and analyze their emotions. When an image is provided, the CNN model identifies facial regions, outlines them with rectangles, and assesses the corresponding emotions. Based on this analysis, a suitable emoji is displayed to represent the detected emotion. This approach brings a more nuanced understanding of visual sentiment by concentrating on human expressions, providing a more intuitive depiction of the emotional content in images. With this work, visual sentiment enhancement analysis can be by emphasized the role of facial emotions, leading to more accurate interpretations and potential applications in various domains.