Age Estimation Using Multi-Label Learning

Generally, age estimation is formulated as a single-label based problem. However, since aging is a gradual process and people are always in transition period between ages, labeling a facial example with an exact age is a difficult problem. Meanwhile, sufficient training data is lack for many ages. In this paper, to improve the accuracy of age estimation, we propose a novel approach by applying Multi-Label Learning to the age features. In the proposed approach, each facial image is treated as an example associated with the origin label as well as its neighboring ages, which makes the data more reliable and sufficient. The motivation comes from the observation that, with age changes slowly and smoothly, people would look quite like themselves before and after several years. Experiments show that the proposed approach outperforms the traditional age estimation approaches.