A Cross-Dimension Annotations Method for 3D Structural Facial Landmark Extraction Academic Article uri icon

abstract

  • AbstractRecent methods for 2D facial landmark localization perform well on closetofrontal faces, but 2D landmarks are insufficient to represent 3D structure of a facial shape. For applications that require better accuracy, such as facial motion capture and 3D shape recovery, 3DA2D (2D Projections of 3D Facial Annotations) is preferred. Inferring the 3D structure from a single image is an illposed problem whose accuracy and robustness are not always guaranteed. This paper aims to solve accurate 2D facial landmark localization and the transformation between 2D and 3DA2D landmarks. One way to increase the accuracy is to input more precisely annotated facial images. The traditional cascaded regressions cannot effectively handle large or noisy training data sets. In this paper, we propose a MiniBatch Cascaded Regressions (MBCR) method that can iteratively train a robust model from a large data set. Benefiting from the incremental learning strategy and a small learning rate, MBCR is robust to noise in training data. We also propose a new CrossDimension Annotations Conversion (CDAC) method to map facial landmarks from 2D to 3DA2D coordinates and vice versa. The experimental results showed that CDAC combined with MBCR outperforms thestateoftheart methods in 3DA2D facial landmark localization. Moreover, CDAC can run efficiently at up to 110 fps on a 3.4 GHzCPU workstation. Thus, CDAC provides a solution to transform existing 2D alignment methods into 3DA2D ones without slowing down the speed. Training and testing code as well as the data set can be downloaded from https://github.com/SWJTU3DVision/CDAC.

published proceedings

  • COMPUTER GRAPHICS FORUM

author list (cited authors)

  • Gong, X., Chen, P., Zhang, Z., Chen, K. e., Xiang, Y., & Li, X.

citation count

  • 0

complete list of authors

  • Gong, Xun||Chen, Ping||Zhang, Zhemin||Chen, Ke||Xiang, Yue||Li, Xin

publication date

  • February 2020

publisher