This paper proposes a new face model named “Optimal Gradient Pursuit Model”, where the objective is to minimize the angle between the gradient direction and the vector pointing toward the ground-truth shape parameter.
Face alignment aims to fit a deformable landmark-based mesh to a facial image so that all facial features can be located accurately. In discriminative face alignment, an alignment score function, which is treated as the appearance model, is learned such that moving along its gradient direction can improve the alignment. The authors formulate an iterative approach to solve this minimization problem. With extensive experiments in generic face alignment, the authors show that their model improves the alignment accuracy and speed compared to the state-of-the-art discriminative face alignment approach. (Publisher abstract provided)