Face Alignment using a 3D Deeply-initialized Ensemble of Regression Trees
3DDE is a robust face alignment algorithm that combines CNN features with an Ensemble of Regression Trees (ERT). It uses a 3D model fitted to CNN probability maps for initialization, then employs a coarse-to-fine cascade of ERT regressors to estimate non-rigid deformations. This hybrid approach handles occlusions and pose variations effectively, achieving state-of-the-art results across multiple datasets.