3D Facial Landmark Detection Using Deep Convolutional Neural Networks

Various applications have been developed in facial research to aid the recognition of personal attributes, analysis of race, and personal authentication for the security industry and other research fields. Thus, it is possible to identify the differences in facial shape based on place of birth or country. The present study analyzes facial shape using scanned 3D facial images and investigates ways to extract facial landmarks from the 3D facial images. The detection of the facial landmark requires the normalization of the facial scale and position among in the 3D image data to analyze the facial shape. Therefore, it is difficult to obtain accurate facial landmarks from 3D facial images. Our method decomposes the task into the following three parts: (a) conversion of data from the 3D facial image to a 2D image, (b) extraction of facial landmarks from the 3D image using Convolutional Neural Network (CNN), (c) inversion of the identified facial landmarks from 2D to 3D images. In experiments, we compared the accuracy of this model for facial landmark detection.