Facial Image-Based Emotion Recognition Using Deep Learning Techniques

Dhanshri Arvind Shrikhande, Dr. Ayesha Siddiqui

Abstract: The detection of motions via facial image processing is one of the most critical areas in artificial intelligence, which concentrates on recognizing emotions from human facial expressions. People tend to reveal their emotions such as happiness, sadness, anger, surprise, fear, and disgust via changes in their facial features. The purpose of this project is to create a system which will be capable of recognizing and classifying those emotions using the captured human facial images. The first step for this task will be the detection of key features of human face images, including eye movements, the shape of the eyes and eyebrows, mouth positions, etc. Afterwards, machine learning and deep learning methods will be applied in order to analyze the patterns of detected features and make appropriate classification of the displayed emotions. Typically, Convolutional Neural Networks (CNNs) are used because of their outstanding efficiency in image recognition tasks. There are plenty of potential applications for such type of technology in different fields. They may be used in healthcare to control patients' mental state, in education to improve the experience of online classes, in security, and many others. In general, emotion detection via facial image analysis facilitates human-machine interactions significantly. Facial-based emotion detection has become an evolving research topic in computer vision and artificial intelligence concerned with recognizing emotions expressed by humans. In this project, we propose a system for automatic detection and classification of emotions based on input images by applying deep learning algorithms. First, the system will detect an input image and use face detection technology to detect the face in the image, after which relevant facial feature extraction will be performed before passing them through the Convolutional Neural Network (CNN). A CNN model will be developed that will have learned the unique patterns of different facial emotions such as happiness, sad, anger, surprise, fear, and neutral faces. After learning the relevant facial features, emotion classification will be performed, and results obtained.

Keywords: Emotion Recognition from Facial Expressions, Artificial Intelligence, Deep Learning, Convolutional Neural Networks (CNNs), VGG-16, Image Processing, Feature Extraction, Emotion Recognition, Transfer Learning, Computer Vision, Face Detection, Machine Learning, Human-Computer Interaction, Multimodal Emotion Recognition, Real-time Emotion Detection