Detection and Classification of Human Emotion Recognition using CNN Algorithm

Facial expressions are hints and signals that help convey messages, share thoughts and emotions, and express reactions through movements of the face. Recognizing human emotions is a challenging task because different people express the same emotions in distinct ways due to cultural differences and personal habits. Deep learning architectures like Convolutional Neural Networks (CNN) provide exceptionally better results for facial detection than traditional methodologies using image processing. Our essential focus is to explore the capability of deep learning techniques, specifically using the traditional CNN architecture, where we examined the resilient FER2013 dataset. In our examination, we carefully evaluated the performance of our model by using various parameters to detect and classify emotions into seven basic categories (happy, sad, angry, fear, disgust, neutral, surprise). We aspire to explore how the image is processed in the input layer and how its final layer predicts the emotional state. We plan to use CNN to achieve greater accuracy in the estimates so that, in further applications, we can develop real-time applications with high accuracy.