Learning from Longitudinal Face Demonstration - Where Tractable Deep Modeling Meets Inverse Reinforcement Learning
This paper presents a novel Generative Probabilistic Modeling under an Inverse Reinforcement Learning approach, named Subject-dependent Deep Aging Path (SDAP), to model the facial structures and the longitudinal face aging process of given subjects. The proposed SDAP is optimized using tractable log-likelihood objective functions with Convolutional Neural Networks based deep feature extraction. In addition, instead of using a fixed aging development path for all input faces and subjects, SDAP is able to provide the most appropriate aging development path for each subject that optimizes the reward aging formulation. Unlike previous methods that can take only one image as the input, SDAP allows multiple images as inputs, i.e. all information of a subject at either the same or different ages, to produce the optimal aging path for the subject. Finally, SDAP allows efficiently synthesizing in-the-wild aging faces without a complicated pre-processing step. The proposed method is experimented in both tasks of face aging synthesis and cross-age face verification. The experimental results consistently show the state-of-the-art performance using SDAP on numerous face aging databases, i.e. FG-NET, MORPH, AginG Faces in the Wild (AGFW), and Cross-Age Celebrity Dataset (CACD). The method also performs on the large-scale Megaface challenge 1 to demonstrate the advantages of the proposed solution.
READ FULL TEXT