Age estimation by extracting hierarchical age-related features

Image-based facial age estimation is considered an intractable problem because aging characteristics are hard to obtain. Most previous works have focused on extracting age-related features, but rarely explored which local region plays an important role. Several works combine local face regions with global face to estimate age in a heuristic way, where the local regions are uniformly cropped for each individual. In this paper, we design an individual adaptive segmentation of local regions of interest to perform personalized local features extraction and build hierarchical age features by erasing the local regions of interest iteratively for each individual. A joint multi-input and multi-output (MIMO) network for multi-task learning of age classification and regression tasks is designed by combining global features and personalized local features as inputs. In addition, we conduct extensive experiments to validate the effectiveness of the proposed method for age estimation, which beats most state-of-the-art methods in three public datasets and also works well for gender and race estimation.

Automated summary

In this paper, a method is proposed that performs personalized local feature extraction and builds hierarchical age features by iteratively erasing local regions of interest. A joint multi-input and multi-output network is designed to learn age classification and regression tasks, using global features and personalized local features as inputs. Extensive experiments validate the effectiveness of the proposed method for age estimation.