Enhancing Facial Emotion Detection with CNN: Exploring the Impact of Hyperparameters

Research on Facial Emotion Recognition (FER) in computer vision and machine learning has significantly increased in the last few years. Effective communication and social interaction greatly depend on one’s capacity to recognize and understand emotions from facial expressions. As a result, FER has attracted a lot of interest from academics in a variety of fields, such as psychology, HCI, and security. In particular, the use of deep learning techniques for FER is the main focus of our work, with a particular emphasis on the evaluation of convolutional neural networks (CNNs). CNNs are chosen for their outstanding performance in image classification tasks as well as their ability to automatically extract crucial characteristics from pictures. To train expression recognition algorithms, this work requires considerable investigation and analysis of different datasets. Notably, our study improves the accuracy of the cutting-edge FER model to 98.85%, outperforming previous models. This work not only adds to our understanding of facial emotion detection and identification, but it also provides insight on the elements that influence its success.