Fusing Deep Learning and Ensemble Techniques for Comic Character Emotion Recognition

Emotions play an essential part in human communication and are tightly entwined with visual storytelling techniques such as comic panel visuals. This research studies the use of Convolutional Neural Networks (CNN) in conjunction with the Random Forest (RF) algorithm for the purpose of accurately classifying the range of emotions that may be found in comic book panels. A carefully selected dataset of 10,000 comic panel images was collected, and it included a spectrum of emotions such as happiness, sadness, rage, surprise, and disgust. The dataset was comprised of carefully picked comic panel images. The hybrid CNN-RF model achieved a great overall accuracy rate of 94.6% after making use of careful preprocessing approaches and carrying out feature extraction. The hybrid technique that was given in this work not only improves the classification of emotions that are depicted in comics, but it also broadens its potential applications in the fields of human-computer interaction, sentiment analysis, and content recommendation systems. The findings highlight the potential for combining art and technology as a tool to convey the many emotional facets that are present in visual narratives. This research presents a vital contribution to the current body of knowledge on emotions within the framework of storytelling, signifying a notable improvement in the field of emotion classification, and it does so by focusing on how emotions are evoked through the medium of narrative.