FACIAL EXPRESSION RECOGNITION SYSTEM USING DEEP LEARNING MODELS BASED ON HUMAN EMOTIONS THROUGH CLASSIFICATION WITH CNN,RNN AND YOLO OBJECT DETECTION ALGORITHMS

Abstract

Emotion is a key topic in a variety of professions, including biomedical engineering,
psychology, neuroscience, and mental health. This component of emotion recognition
is crucial, because it is commonly used in the diagnosis of human brain and
psychiatric diseases. Deep learning has gotten a lot of users' interest in the field of
picture categorization, according to a recent poll. These emotions are employed not
just for brain diagnosis, but also as a recommendation system to help consumers
select goods that meet their requirements and preferences. This inspired us to create a
system that can accurately and efficiently discern emotions based on the user's facial
expressions. In this proposal, we aim to create an application that can be used to
anticipate expressions in both still and moving photographs. Then compare the results
of the CNN with the recurrent neural network (RNN) model. Once the image is taken
from the video sequences, the system uses HAAR cascade to detect faces, crops the
image, resizes it to the necessary dimension, and sends it to the model for prediction.
Seven probability values will be generated by the model, matching to seven
expressions. We compare the two models to see which one provides better face
expression detection accuracy for the image dataset.