Classificação de emoções usando inteligência artificial para maior segurança no trânsito

Abstract

The development of smart cars is a promising technology for the 21st century, with safety being a key concern. In this context, the incorporation of advanced Artificial Intelligenceand computer vision techniques, such as emotion recognition, emerges as a potentialsolution to enhance traffic safety by enabling driver analysis and preventing dangerousbehaviors. Therefore, the objective of this work is to utilize a convolutional neural network (CNN) for emotion classification in videos, aiming at traffic safety. Facial analysis is used to extract relevant features for classification. The results obtained with the proposed methodology demonstrated an average precision of 82.82%, average sensitivity of 80.87%, and average accuracy of 81.00% in emotion classification. These results reinforce the importance and potential of emotion recognition techniques to improve traffic safety and contribute to future studies in the field.