DeepI2I: Enabling Deep Hierarchical Image-to-Image Translation by Transferring from GANs

TL;DR

DeepI2I introduces a hierarchical image-to-image translation approach that leverages transfer learning from pre-trained GANs, significantly improving performance on small datasets and enabling translation across many classes.

Contribution

The paper proposes a novel deep hierarchical I2I method combined with transfer learning from GANs, addressing high-resolution bottlenecks and enabling multi-class translation.

Findings

Decreased mFID by at least 35% compared to state-of-the-art.

Transfer learning significantly boosts I2I performance on small datasets.

First to perform I2I translation across over 100 classes.

Abstract

Image-to-image translation has recently achieved remarkable results. But despite current success, it suffers from inferior performance when translations between classes require large shape changes. We attribute this to the high-resolution bottlenecks which are used by current state-of-the-art image-to-image methods. Therefore, in this work, we propose a novel deep hierarchical Image-to-Image Translation method, called DeepI2I. We learn a model by leveraging hierarchical features: (a) structural information contained in the shallow layers and (b) semantic information extracted from the deep layers. To enable the training of deep I2I models on small datasets, we propose a novel transfer learning method, that transfers knowledge from pre-trained GANs. Specifically, we leverage the discriminator of a pre-trained GANs (i.e. BigGAN or StyleGAN) to initialize both the encoder and the discriminator and the pre-trained generator to initialize the generator of our model. Applying knowledge transfer leads to an alignment problem between the encoder and generator. We introduce an adaptor network to address this. On many-class image-to-image translation on three datasets (Animal faces, Birds, and Foods) we decrease mFID by at least 35% when compared to the state-of-the-art. Furthermore, we qualitatively and quantitatively demonstrate that transfer learning significantly improves the performance of I2I systems, especially for small datasets. Finally, we are the first to perform I2I translations for domains with over 100 classes.