Better initialization for regression-based face alignment. (February 2018)

• Record Type: Journal Article
• Authors: Zhu, Hengliang; Sheng, Bin; Shao, Zhiwen; Hao, Yangyang; Hou, Xiaonan; Ma, Lizhuang
• Is Part Of: Computers & graphics. Volume 70(2018)
• Page Start: 261
• Page End: 269
• Publication Date: 2018-02
• Subjects: Neighborhood representation prior -- Occlusions -- Projected initial shape -- Cascade regression; Computer graphics -- Periodicals; 006.6
• DOI: 10.1016/j.cag.2017.07.036
• Languages: English
• ISSNs: 0097-8493

Abstract

Highlights: We propose a neighborhood representation prior to generate a high-quality initial shape. We propose a face-like landmark adjustment algorithm to refine the face shape. In the condition of complex scenarios, our algorithm can significantly promote the accuracy of the cascade regression-based face alignment. Our method achieves the performance of about 223 fps, and it can be used to real-time facial landmark tracking. Graphical abstract: Abstract: Regression-based face alignment algorithms predict facial landmarks by iteratively updating an initial shape, and hence are always limited by the initialization. Usually, the initial shape is obtained from the average face or by randomly picking a face from the training set. In this study, we discuss how to improve initialization by studying a neighborhood representation prior, leveraging neighboring faces to obtain a high-quality initial shape. In order to further improve the estimation precision of each facial landmark, we propose a face-like landmark adjustment algorithm to refine the face shape. Extensive experiments demonstrate our algorithm achieves favorable results compared to the state-of-the-art algorithms. Moreover, our algorithm achieves a smaller normalized mean error than the human performance (5.54% vs. 5.6%) on the challenging dataset the Caltech Occluded Faces in the Wild (COFW).