Hailin Li defended his Master Thesis on Face Age Progression with Realness Distributions
Face age progression was a newly discussed topic in recent years, which can be used in entertainment purpose photography effect. It can even be contributed to the social service department such as assist police to simulate the future appearances of a wanted criminal
or help to find lost children. With the success of the deep learning based method, the face generated task achieved impressive performance through Generative Adversarial Networks. Until now, many state-of-the-art approaches had successfully produced high resolution face images with age effects. The contribution of this thesis is the development of a face age progression model based on an encoder-decoder architecture, training with a high resolution database with 70000 images. Compared to the baseline model, the identity preservation, visual fidelity remains as much as possible but improved the age accuracy extremely.