Unsupervised Discovery of Disentangled Manifolds in GANs
This work addresses the problem of understanding and controlling the generation process in GANs for image editing applications, which is an incremental improvement for practitioners.
This paper proposes a framework to discover interpretable directions in the latent space of pre-trained Generative Adversarial Networks (GANs). By learning transformations from one-hot vectors to the latent space and applying a centroid loss, the framework enables attribute editing through visually distinct attribute vectors.
As recent generative models can generate photo-realistic images, people seek to understand the mechanism behind the generation process. Interpretable generation process is beneficial to various image editing applications. In this work, we propose a framework to discover interpretable directions in the latent space given arbitrary pre-trained generative adversarial networks. We propose to learn the transformation from prior one-hot vectors representing different attributes to the latent space used by pre-trained models. Furthermore, we apply a centroid loss function to improve consistency and smoothness while traversing through different directions. We demonstrate the efficacy of the proposed framework on a wide range of datasets. The discovered direction vectors are shown to be visually corresponding to various distinct attributes and thus enable attribute editing.