Augmentation-Interpolative AutoEncoders for Unsupervised Few-Shot Image Generation
This work provides a simple and lightweight unsupervised method for few-shot image generation, which is significant for researchers and practitioners working on data-scarce domains.
This paper tackles the problem of generating images for new domains with few examples by leveraging the generalization capabilities of autoencoders. The proposed Augmentation-Interpolative AutoEncoders synthesize realistic images from a few references, outperforming prior interpolative and supervised few-shot image generators.
We aim to build image generation models that generalize to new domains from few examples. To this end, we first investigate the generalization properties of classic image generators, and discover that autoencoders generalize extremely well to new domains, even when trained on highly constrained data. We leverage this insight to produce a robust, unsupervised few-shot image generation algorithm, and introduce a novel training procedure based on recovering an image from data augmentations. Our Augmentation-Interpolative AutoEncoders synthesize realistic images of novel objects from only a few reference images, and outperform both prior interpolative models and supervised few-shot image generators. Our procedure is simple and lightweight, generalizes broadly, and requires no category labels or other supervision during training.