Automated Facial Expression Recognition Framework Using Deep Learning.

Area of Science:

• Computer Science
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Facial expressions are crucial for nonverbal communication, conveying significant information about a person's mental state.
• Automatic Facial Expression Recognition (FER) is a challenging computer science task with diverse applications in healthcare, security, and entertainment.
• FER is increasingly vital in medical sciences, aiding in the monitoring of conditions like bipolar disease.

Purpose of the Study:

• To propose an automated framework for facial detection using a convolutional neural network (FD-CNN) to improve FER accuracy.
• To enhance the recognition of various facial expressions including anger, fear, disgust, and happiness.

Main Methods:

• An FD-CNN algorithm with four convolution layers and two hidden layers was developed.
• The extended Cohn-Kanade (CK+) dataset, comprising diverse facial expressions, was utilized.
• The methodology involved three key steps: preprocessing, feature extraction, and classification.

Main Results:

• The proposed FD-CNN method achieved an accuracy of 94% for facial expression recognition.
• K-fold cross-validation confirmed the algorithm's robustness.
• High performance metrics were reported: sensitivity (94.02%), specificity (99.14%), F1 score (84.07%), recall (78.22%), and precision (94.09%).

Conclusions:

• The developed FD-CNN framework demonstrates high accuracy and reliability for facial expression recognition.
• The study validates the effectiveness of deep learning approaches in analyzing complex human emotions.
• The proposed FER system offers potential for advancements in various fields, including mental health monitoring and human-computer interaction.