Improved optimizer with deep learning model for emotion detection and classification.

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Facial emotion recognition (FER) is crucial for applications like human-computer interaction and biometrics.
• Current FER methods often struggle with accuracy and high error rates.
• There is a need for robust and precise automated facial emotion detection systems.

Purpose of the Study:

• To develop an innovative deep learning framework, the extended walrus-based deep learning with Botox feature selection network (EWDL-BFSN), for accurate facial emotion recognition.
• To automatically identify facial emotions by optimizing feature selection and classifier hyperparameters.
• To enhance the performance of FER systems beyond current state-of-the-art capabilities.

Main Methods:

• The EWDL-BFSN framework integrates gradient wavelet anisotropic filter (GWAF) for image pre-processing and SqueezeNet for feature extraction.
• The improved Botox optimization algorithm (IBoA) is employed for optimal feature selection.
• An enhanced optimization-based kernel residual 50 (EK-ResNet50) network performs FER and classification, with hyperparameters tuned by the walrus optimization algorithm (WOA).

Main Results:

• The EWDL-BFSN model achieved high accuracy rates of 99.37% on the CK+ dataset and 99.25% on the FER-2013 dataset.
• Performance metrics including accuracy, sensitivity, specificity, and F1-score were analyzed.
• The proposed model demonstrated superior performance compared to existing state-of-the-art methods in facial emotion prediction.

Conclusions:

• The EWDL-BFSN framework offers a highly accurate and effective solution for facial emotion recognition.
• The integration of advanced deep learning techniques and optimization algorithms significantly advances FER capabilities.
• The model's superior performance validates its potential for real-world applications requiring precise emotion detection.