Hybrid Grey Wolf Optimizer with Ant Lion Optimization Algorithm based Deep Learning for Face Expression Recognition

Facial expressions are an important part of recognizing human emotional messages, hence it has been a focus of pattern recognition research. However, developments in convolutional neural networks and network topologies have resulted in difficulties such as decreased overall performance and the potential loss of crucial feature information in these methods. To overcome this issue, the hybrid optimization algorithm of Grey Wolf Optimization algorithm (GWO) based Ant Lion Optimization Algorithm (ALO) termed as GWALO is proposed with deep learning for recognizing facial expressions. Viola Jones Algorithm and Gabor Filtering are used in preprocessing to improve image quality by minimizing redundant distortion, allowing for the extraction of relevant information. The Affine-Scale-Invariant Feature Transform is used in feature extraction to extract expressions from a preprocessed image. The hybrid GWALO is utilized in the feature selection for selecting the better features. Recurrent Neural Network (RNN) of Deep Learning (DL) is utilized in classification for classifying the various facial expressions (happiness, joy, sorrow, angry, sad and so on). The obtained result shows that the proposed GWALO algorithm attains accuracy of 99.7% precision of 98.5%, recall of 97.8%, specificity of 99.6% and f1score of 96.7% that is better when contrasted to existing optimization algorithms.