Comparison of YOLO-v8 and YOLO-v10 in Detecting Human Facial Emotions

This study evaluates the performance of YOLOv8 and YOLOv10 in recognizing human facial emotions. Both state-of-the-art object detection models were trained on a diverse dataset of facial expressions. While YOLOv10 demonstrated superior performance in certain metrics, it required significantly more training time compared to YOLOv8. Both models exhibited effective learning, as evidenced by the steady decrease in training loss. However, both models encountered challenges in accurately recognizing subtle emotions, such as disgust and contempt. To enhance the accuracy and robustness of facial emotion recognition systems, future research should prioritize improving data quality, exploring advanced model architectures, and optimizing hyperparameters.