Exploring Unlabeled Faces for Novel Attribute Discovery
This addresses the bottleneck of labeled data dependency for real-world applications in facial image translation, enabling use with unlabeled images, though it is incremental as it builds on existing translation systems.
The paper tackles the problem of requiring large labeled datasets for unpaired image-to-image translation in facial images by proposing a method to discover novel attributes from unlabeled faces and perform high-quality translation, achieving results comparable to or better than state-of-the-art methods trained on labeled data.
Despite remarkable success in unpaired image-to-image translation, existing systems still require a large amount of labeled images. This is a bottleneck for their real-world applications; in practice, a model trained on labeled CelebA dataset does not work well for test images from a different distribution -- greatly limiting their application to unlabeled images of a much larger quantity. In this paper, we attempt to alleviate this necessity for labeled data in the facial image translation domain. We aim to explore the degree to which you can discover novel attributes from unlabeled faces and perform high-quality translation. To this end, we use prior knowledge about the visual world as guidance to discover novel attributes and transfer them via a novel normalization method. Experiments show that our method trained on unlabeled data produces high-quality translations, preserves identity, and be perceptually realistic as good as, or better than, state-of-the-art methods trained on labeled data.