StyleAutoEncoder for manipulating image attributes using pre-trained StyleGAN
This provides a cost-effective solution for image attribute manipulation with limited computational resources, though it is incremental as it builds on existing pre-trained models.
The paper tackles the problem of expensive training for deep conditional generative models by introducing StyleAutoEncoder (StyleAE), a lightweight AutoEncoder plugin for pre-trained models like StyleGAN, which effectively manipulates image attributes with performance comparable to state-of-the-art invertible normalizing flow algorithms.
Deep conditional generative models are excellent tools for creating high-quality images and editing their attributes. However, training modern generative models from scratch is very expensive and requires large computational resources. In this paper, we introduce StyleAutoEncoder (StyleAE), a lightweight AutoEncoder module, which works as a plugin for pre-trained generative models and allows for manipulating the requested attributes of images. The proposed method offers a cost-effective solution for training deep generative models with limited computational resources, making it a promising technique for a wide range of applications. We evaluate StyleAutoEncoder by combining it with StyleGAN, which is currently one of the top generative models. Our experiments demonstrate that StyleAutoEncoder is at least as effective in manipulating image attributes as the state-of-the-art algorithms based on invertible normalizing flows. However, it is simpler, faster, and gives more freedom in designing neural