Rethinking the Truly Unsupervised Image-to-Image Translation

TL;DR

This paper introduces TUNIT, a fully unsupervised image-to-image translation model that learns to separate and translate image domains without requiring paired data or domain labels, outperforming some supervised methods.

Contribution

We propose TUNIT, the first model for fully unsupervised image-to-image translation that does not rely on paired images or domain labels, and demonstrate its effectiveness.

Findings

Achieves comparable or better performance than supervised models.

Generalizes well across various datasets.

Robust to hyperparameter choices.

Abstract

Every recent image-to-image translation model inherently requires either image-level (i.e. input-output pairs) or set-level (i.e. domain labels) supervision. However, even set-level supervision can be a severe bottleneck for data collection in practice. In this paper, we tackle image-to-image translation in a fully unsupervised setting, i.e., neither paired images nor domain labels. To this end, we propose a truly unsupervised image-to-image translation model (TUNIT) that simultaneously learns to separate image domains and translates input images into the estimated domains. Experimental results show that our model achieves comparable or even better performance than the set-level supervised model trained with full labels, generalizes well on various datasets, and is robust against the choice of hyperparameters (e.g. the preset number of pseudo domains). Furthermore, TUNIT can be easily extended to semi-supervised learning with a few labeled data.