Expression Recognition Survey Through Multi-Modal Data Analytics

In computer vision, capturing human expression from a video is essential for various time-sensitive applications, including driver safety and education. Because there are multiple expression models (e.g., speech, face, gesture, etc. ), Human Expression Recognition can be done in various methods. In this work, we look at multiple strategies for recognizing facial expressions. We primarily concentrated on two models in this review: speech signal and facial image. The entrance survey is conducted in two steps, according to the generic approach of human expression recognition.The methods we used to extract features were classified into two groups: audio features and facial features. The review found different classifiers helped to decide the best framing of the Markov chain for a model. We also looked at specific multimodal expression recognition algorithms that used the multimodal notion at various stages, such as feature fusion and decision fusion. We also looked at several databases that have been used by previous studies in addition to these approaches. A full-fledged comparison is also provided to show the review ultimately.