Systemic Review and Design of Facial Emotion Recognition System using Ensemble ML

During the COVID era, this study describes a facial expression detection system to identify emotions connected to depression. The process entails collecting a collection of depression-related photos and videos, manually labelling emotions, and detecting faces using a Viola-Jones object detection framework. Key landmarks such as the eyes, nose, and mouth are extracted using facial feature detection algorithms. The identified features are used to compute relevant facial measurements such as eye-to-eye distance and nose width, and facial feature vectors are formed by combining measurements with emotion labels. For emotion recognition, an ensemble learning strategy with kernel-based Self-Organizing Maps (SOM) is applied. Model evaluation and hyperparameter fine-tuning are carried out to increase performance. The created system recognises emotions from photos and videos, with performance metrics based on accuracy and mean delay. Furthermore, the fundamentals of depression and the use of machine learning, namely MATLAB, are presented. The method provides vital insights into individuals’ emotional states throughout the COVID era and aids in the treatment of mental health issues.