A face reconstruction method based on fusion regression network and gradient descent

Menglin Zhao; Ming Yan

doi:10.1117/12.2542617

27 November 2019 A face reconstruction method based on fusion regression network and gradient descent

Menglin Zhao, Ming Yan

Proceedings Volume 11321, 2019 International Conference on Image and Video Processing, and Artificial Intelligence; 1132121 (2019) https://doi.org/10.1117/12.2542617
Event: The Second International Conference on Image, Video Processing and Artifical Intelligence, 2019, Shanghai, China

Abstract

Three-dimensional (3D) face reconstruction refers to the restoration and reconstruction of 3D model of face from one or more two-dimensional (2D) images. It has been widely used in face recognition, expression migration, face editing and other aspects. In the current existing algorithms, there are still many shortcomings in how to reconstruct 3D face by parametric model in real time. In this paper, based on the convolutional neural network, we integrate the weight mask into the loss function, and then use the back propagation algorithm to calculate the parameter gradient error. Finally, the parameter self-renewal purpose of the loss function is achieved by gradient descent. It can be seen from the experimental results that this method can accurately reconstruct the 3D contour of the face, and the reconstruction results are complete and the topological structure is known. This is very important for the application after face reconstruction, such as face changing, expression changing and other aspects of accuracy has been greatly improved.

Citation Download Citation

Menglin Zhao and Ming Yan "A face reconstruction method based on fusion regression network and gradient descent", Proc. SPIE 11321, 2019 International Conference on Image and Video Processing, and Artificial Intelligence, 1132121 (27 November 2019); https://doi.org/10.1117/12.2542617

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available