UNSUPERVISED ADAPTATION OF A PERSON-SPECIFIC MANIFOLD OF FACIAL EXPRESSION

Abstract

The project offers feature analysis and NN-classifiers-based speech emotion identification from voice data. By associating facial expressions with categories of fundamental emotions, automatic face emotion recognition (SER) has dominated psychology and plays a significant role in HCI systems for evaluating people's emotions (i.e., anger, disgust, fear, happiness, sadness, and surprise). Facial emotion detection, feature extraction and selection, and classification are all parts of the recognition system. These characteristics are helpful in properly differentiating the greatest number of samples, and the discriminant analysis-based NN classifier is utilised to categorise the six distinct phrases. The simulated results will demonstrate that, compared to previous Face emotion expression recognition systems, filter-based feature extraction with a utilised classifier provides substantially greater accuracy with less algorithmic complexity.