Face Deidentification and Morphing using Deep Learning

Privacy protection has become a major concern in today's digital world.Particularly sensitive data such as facial images, which not only reveal a person's identity, but also other sensitive personal information.Nowadays hackers are trying to collect the sensitive data of people and sell that information to the right person who will in-turn use that information for malicious activity.Any individual does not want their face features to be collected by the other persons with out their concern.For instance, if a hacker obtains a person's facial dataset, they could exploit it to monitor the individual's activities.This might involve seizing control of the CCTV system in the person's office building to observe their movements.Alternatively, the hacker could crossreference the dataset with public databases to extract sensitive information such as PAN card and Aadhar card numbers.To mitigate these risks, we've created a model that anonymizes facial features, providing protection against detection by facial recognition models.In this face de-identification we are using Delauny Triangulation, landmark detection algorithms.Morphing, entails the transformation of facial characteristics to create a synthetic identity.This process aims to generate visual content that is not directly linked to any real individual, thereby mitigating the risk of unauthorized identification.We suggest employing a morphing-based approach to safeguard visual privacy, as we believe it encompasses all the essential qualities of an effective visual privacy protection method without significant drawbacks.