FACIAL ATTRIBUTE EDITING USING GANs FOR CRIMINAL INVESTIGATION SYSTEM

Abstract

Facial attributes are significant for the identification of person. The facial attributes changes with time and desire of the person. But facial attributes remain the same in photographs or images. Manipulation of facial attributes in a digital image is widely used for entertainment purposes but its scope is way more than that. Most of the facial attribute editing tools uses image processing which consume more processing power. The evolving AI technology has made this task more cheaper in terms of computation. This project emphasizes on facial attribute editing using AI. The core of this project is the GAN powered face attribute editing. GANs are capable of generating new images with adversarial training. This project makes use of AttGAN architecture, which generates the realistic face image of a person with the customised attribute set. Unlike traditional approach of overwriting the existing image, this approach generates completely new image. This project has achieved to put on mask along with other 27 facial attributes to the image. This has broadened the application including its use cases for criminal investigation. Changing the facial attributes for disguise, estimating the face of long term missing child comes under criminal investigation where it can be used.