Hybrid Optimization-based Remodel-Generative Adversarial Network for Face Detection and Reconstruction of Face Region Behind Face Mask

Due to COVID-19, facial coverings like masks have become increasingly widespread. Although enhanced facial recognition approaches have been introduced and create reasonable outcomes, obstructions and covering of facial features are the prime concern. As a mask is one of the things that should be categorized as an obstacle to a face, research has been developed that removes the mask and generates or reconstructs the hidden part of the face covered by the mask utilizing Jellyfish Search Eel and Grouper Optimization-based Remodel-Generative Adversarial Network (JSEGO_Re-GAN). First, an input is pre-processed by employing a Gaussian filter and then Region of Interest (ROI) extraction of the face region is carried out. Afterwards, face mask detection is conducted through Deep Fuzzy Clustering (DFC), which is trained by the Adaptive Addax Optimization Algorithm (Adaptive AOA). Then, the face region behind the face mask is reconstructed using Re-GAN, where the hyperparameters of Re-GAN are tuned by JSEGO. Here, Re-GAN is designed by modifying GAN’s layers. The JSEGO_Re-GAN has attained the analytic measures of Normalized Mean Square Error (MSE), Normalized Root Mean Square Error (RMSE) and Normalized [Formula: see text]-square as 0.343, 0.585 and 0.364.