An Efficient Emotion Recognition Model Using Bald Hawk Optimization based Activation Attention Deep CNN Classifier

Abstract

The enormous variety of facial expressions makes it challenging to determine emotions from face images. Emotion detection was the main focus of earlier research on detecting emotions from facial pictures using deep learning models, but there is an issue with performance degradation caused by poor layer selection as well as poor accuracy. In order to accurately classify emotions from facial images, a bald hawk optimized activation attention module deep CNN (BHO-deep CNN) technique was developed in this research to address this problem.In the beginning, data is gathered, and preprocessing is done to improve the image's quality for more accurate emotion recognition. The region of extraction is then carried out using faster R-CNN hybridization and the standard Resnet-101 hybridization. Then, a hybridized pre-trained model created using the standard hybridization of the Resnet-101 and Google net model was used for feature extraction. Once the features have been collected, a deep CNN classifier with an optimized attention module is used to quickly categories the emotions visible in the image.The classifier's weights and bias are successfully adjusted by the BHO method. The outputs' accuracy, sensitivity, and specificity are evaluated. In comparison to existing methods, the suggested BHO-deep CNN achieves values of 94.01%, 95.43%, 94.47% for dataset 1 and 94.78%, 94.88%, 92.72% for dataset 2, which is more efficient than other techniques.