Improved Unconstrained Face Recognition Quality By Using Reference Face Graph

Abstract

: A multitude of applications in modern digital era, including security, surveillance, biometric authentication, and human-computer interface, have fueled the demand for accurate and dependable face recognition systems. Using state of heart computational tools, this research aims to investigate novel approaches to improve unconstrained face recognition in varied and dynamic settings. For situations involving unconstrained face recognition, the first method dives into edge detection using the Side Searching Method SSM and the Object Improving Method OIM to improve picture quality. Then, to improve recognition accuracy and scalability, a new reference face based method is suggested, which uses a reference face graph RFG. With its solid foundation for handling the difficulties of real world face recognition tasks, this approach is a huge step forward in the area. In addition, the study uses a hybrid estimating method to successfully extract and recreate 3D facial features, while tackling the complex issues of unconstrained face feature recognition in video streams. This method has the potential to completely change the face recognition game by making it more accurate and flexible in real-time video settings. Finally, synthetic face images are generated by integrating Deep Convolutional Generative Adversarial Networks DCGANs, which improves face graph representations and strengthens resistance against real world changes. The project aims to advance facial recognition technology using various approaches, providing potential solutions to meet the changing needs of modern society. newline