Towards 3D facial reconstruction using deep neural networks

Abstract

3D facial reconstruction is an emerging and interesting application in the field of computer graphics and computer vision and is been used for 3D printing, creating avatars etc. It is difficult and challenging to reconstruct the 3D model of face from a single photo because of arbitrary poses, non-uniform illumination, expressions and occlusions. None of algorithm provides detailed 3D facial model because every algorithm has some limitations related to profile view, fine detail, accuracy and speed. The major problem is to develop 3D face with texture of large poses, wild faces and occluded faces. Mostly algorithm is used convolution neural network (CNN) and deep learning frameworks to create facial model and 3D dense face alignment (3DDFA) is the first algorithm that constructed the database consisting of 2D images and 3D facial model. In this paper we review 3D face reconstruction algorithm for application such as 3D printing, creating avatars and facial recognition. Different issues, problems and their proposed solutions are discussed in this paper while advantages and disadvantages are highlighted. A comparison of different algorithms is described in the context of texture and poses to find the best solution regarding reconstruction of 3D facial model from single photo.