Optimizing Real-Time Emotion Recognition: A YOLO v.8 Deep Learning Solution for Facial Expression Analysis

Identifying emotional states from facial expressions is challenging and requires advanced technology. This paper introduces a YOLO (you only look once) based-deep learning facial expression recognition model. This rapid and precise object detection method improves facial recognition. The project aims to comprehend face recognition principles by assessing Yolo v.5, Yolo v.8s (Small), and Yolo v.8m (Medium) based models' facial expression recognition. Moreover, evaluate system performance using accuracy, recall, F1-score, precision, and robustness; measures Yolo v.8m is precise and efficient. The model is practical due to its accuracy and efficiency across several criteria. Yolo v.5 is weaker to other models but accurately catches face emotions with 91% and 0.87 macro-average F1-score. At 93% accuracy and 0.86, Yolo v.8s has good recall, precision, and macroaverage F1-score. Yolo v.8m balanced F1-score and 98% accuracy make it the best of three models. Its balanced macro-average F1-score i.e., 0.88. Potentially revolutionary, the suggested approach offers a solid fix for practical problems where traditional approaches fall short.