Few-shot Knowledge Transfer for Fine-grained Cartoon Face Generation
This addresses the challenge of few-shot learning for domain-specific applications like cartoon face generation, though it appears incremental as an adaptation of existing transfer learning methods.
The paper tackles the problem of generating fine-grained cartoon faces for groups with limited training data by transferring knowledge from a well-trained basic group model. The proposed two-stage approach creates group-specific branches for new groups while maintaining shared parameters, enabling high-quality generation across various groups.
In this paper, we are interested in generating fine-grained cartoon faces for various groups. We assume that one of these groups consists of sufficient training data while the others only contain few samples. Although the cartoon faces of these groups share similar style, the appearances in various groups could still have some specific characteristics, which makes them differ from each other. A major challenge of this task is how to transfer knowledge among groups and learn group-specific characteristics with only few samples. In order to solve this problem, we propose a two-stage training process. First, a basic translation model for the basic group (which consists of sufficient data) is trained. Then, given new samples of other groups, we extend the basic model by creating group-specific branches for each new group. Group-specific branches are updated directly to capture specific appearances for each group while the remaining group-shared parameters are updated indirectly to maintain the distribution of intermediate feature space. In this manner, our approach is capable to generate high-quality cartoon faces for various groups.