Optimal Facial Feature Based Emotional Recognition Using Deep Learning Algorithm.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Machine Learning

Background:

• Facial expression recognition is challenging for computers, despite its ease for humans.
• Automatic emotion recognition is crucial for human-computer interaction and understanding mental states.
• Deep learning, particularly Convolutional Neural Networks (CNNs), shows promise for complex recognition tasks.

Purpose of the Study:

• To improve Convolutional Neural Network (CNN) performance for recognizing seven fundamental emotions from facial expressions.
• To evaluate the impact of various preprocessing techniques on CNN-based emotion recognition.
• To enhance the detection of facial features and expressions for more accurate emotion identification.

Main Methods:

• Utilized a deep learning approach employing a Convolutional Neural Network (CNN).
• Trained and tested the model on a dataset of approximately 32,298 images featuring multiple facial expressions.
• Implemented preprocessing techniques for noise removal and feature extraction, including a pretraining phase for face detection.

Main Results:

• The improved CNN technique successfully identified seven fundamental emotions from facial expressions.
• Preprocessing methods were shown to significantly impact CNN performance in emotion recognition.
• The study demonstrated effective classification of facial reactions corresponding to the seven emotions of the Facial Acting Coding System (FACS).

Conclusions:

• The proposed deep learning method using CNNs effectively enhances emotion recognition from facial features.
• Optimized preprocessing and CNN techniques provide a robust solution for identifying basic human emotions.
• This advancement contributes to more accurate predictions of mental states and personalized human-computer interactions.