Frontal Face Landmark Generation using GAN
Avhi Poddar, Surabhi Gawade, Prasad Varpe, Sumedha Bhagwat · 2022 International Conference on Applied Artificial Intelligence and Computing (ICAAIC) · 2022
In the area of video surveillance, there are many challenges identified due to illumination variation, pose variation, angular variation, occlusion etc for face identification. It is very crucial to identify face images from CCTV footages, uncontrolled environment because of low resolution, poor quality of camera and distance between camera and face. This leads to poor quality image generation and difficult to identify face. This paper proposes a novel system which handles pose variation of face. In this paper, GAN (Generative Adversarial Network) is used to generate landmark of frontal face from side view of the face image. 2D face can be generated based on the frontal face landmark we have generated. Experimental results show that frontal face landmark can be generated easily which is further helpful to generate front face.