Privacy Protection Against Reverse Image Search
Rishank Pratik, R. Sendhil · 2023
The user wants to include their image in their résumé, LinkedIn profile, or any other relevant place and doesn't want their personal life to be viewed using a reverse image search, which can trace back their professional image to their social ID based on facial image. They may not want to be evaluated as social media account may expose sensitive information or personal life. Various techniques such as synthetic modification exist to protect the privacy of an individual but they mostly tend to distort the image to such an extent that figure is not identifiable as a human or some other person is represented.This article provides a technique that enables people to protect their images against the misuse of facial recognition software enabling humans to identify the person but machines fail to recognize the person. The proposed method would help the users to edit their images at the pixel level and removing the metadata before sharing them. These "cloaked" images yield functional models that repeatedly lead to incorrect identification of typical images of the user when used to train facial recognition models. When combined with metadata removal, it highly reduces the possibility of image traceback to a person’s social ID.