Abstract:
During the last two decades' security of men and materials remains a vital concern in the whole
world. In particular, Pakistan being a victim of multiple terrorist attacks on military and civil setups
suffered massive damage. To encounter a security threat, a lot of efforts have been done to improve
the organization's security and various enhanced safety checks were incorporated. Moreover, with
the emergence of novel coronavirus pandemic, face masks have become an important part of daily
routine life. Nearly every organization in the world has adopted face masks as a primary
precautionary measure to secure their workplaces. Masked faces have made existing technology
ineffective in several scenarios, such as facial recognition access control and facial security checks
at public places This presents a new challenge to any organization’s security. Since timely
identification of masked faces is vital for an organization. For any setup to ensure right entry at
the gate with an automated system for human face recognition (even with mask) in a real-time
background is the latest requirement. A face recognition system is an application of computer
vision that can perform two tasks identifying and verifying a person from a given database,
intending to reduce the manual efforts of the management and security staff. This research thus
focuses on implementing a face recognition system (even with a face mask) by using a machine
vision-based approach. A dataset of masked faces was collected to train the Support Vector
Machine classifier on state-of-the-art Facial Recognition Feature Extractor Convolution Neural
Network. Proposed Methodology gives recognition accuracy of 98% with masked faces, and
results in an effective gate access control and attendance system.
