An intuitive and effective real time face mask recognition bio metric based attendance system for controlled hospital scenario

Businesses are seeking for solutions to safeguard their employees and clients, from cross-border travel to login and authentication. The transmission of epidemic diseases and possibly other viruses that can be communicated through close contact may be prevented by reducing the number of touch points. A framework must be developed for authentication-based surveillance applications in controlled environments, such as Medical organizations and the IT industry, in order to facilitate security. The issues this system is designed to tackle are position variations, haircuts, poor resolution, goggles, partial occlusions, facial emotions, and identical twin detection. Although deep learning algorithms have gained popularity, they are associated with a large amount of data, which may not be widely available for training. Deep learning is also time consuming, both during training and testing. The present work mitigates all of these drawbacks by relying primarily on handcrafted features. We propose a method for using non-deep learning methods in surveillance. Since the Viola-Jones (VJ) algorithm can identify faces with frontal, mid-profiling, and 35-degree profiles, it is utilised for face detection. The detected faces are then subjected to the proposed fused-multi-block local binary pattern based on Eigen-face and texture. Additionally, the proposed algorithm is capable of recognizing identical twins. In comparison to other sophisticated methods like CS-LBP (79.3%), XCS-LBP (85.7%), MB-LBP (94.6%), and Eigen-face (87.6%), this algorithm has an accuracy of 99.15%.