Sequential Face Alignment via Person-Specific Modeling in the Wild

Xi Peng, Junzhou Huang, Dimitris N. Metaxas; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2016, pp. 107-116

Sequential face alignment, in essence, deals with non-rigid deformation that changes over time. In this paper, we propose to exploit incremental learning for person-specific alignment. Our approach takes advantage of part-based representation and cascade regression for robust and efficient alignment on each frame. More importantly, it incrementally updates the representation subspace and simultaneously adapts the cascade regressors in parallel using a unified framework. Person-specific modeling is eventually achieved on the fly while the drifting issue is significantly alleviated by erroneous detection using both part and holistic descriptors. Extensive experiments on both controlled and in-the-wild datasets demonstrate the superior performance of our approach compared with state of the arts in terms of fitting accuracy and efficiency.