Face alignment based on improved shape searching
Yuqin Huang, Huawei Pan · 2017
Accurate face alignment is a vital step for most face perception tasks. In this paper, we proposed a new approach based on improved face shape searching for face alignment. It begins with a shape space that contains diverse shapes. Unlike previous shape searching method that processes its every shape searching in the whole sample space, our method first train the random forest classifiers by training samples and partition the whole shape space into multiple sub-spaces, and then we process our shape searching in sub-spaces. Finally, we employ cascaded regression to achieve face alignment. Our method demonstrates its obvious decreases in searching time and its good robustness in unconstrained environment on three challenging datasets.