Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

TLDR

Image‑to‑image translation seeks to learn a mapping between input and output images, yet paired training data are often unavailable. The authors aim to learn a mapping from a source domain X to a target domain Y without paired examples. They employ a cycle‑consistent adversarial framework that couples a forward mapping G: X→Y with an inverse mapping F: Y→X, enforcing cycle consistency to constrain the under‑determined problem. Experiments on style transfer, object transfiguration, season transfer, and photo enhancement demonstrate that the method outperforms prior approaches when paired data are absent.

Abstract

Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs. However, for many tasks, paired training data will not be available. We present an approach for learning to translate an image from a source domain $X$ to a target domain $Y$ in the absence of paired examples. Our goal is to learn a mapping $G: X \rightarrow Y$ such that the distribution of images from $G(X)$ is indistinguishable from the distribution $Y$ using an adversarial loss. Because this mapping is highly under-constrained, we couple it with an inverse mapping $F: Y \rightarrow X$ and introduce a cycle consistency loss to push $F(G(X)) \approx X$ (and vice versa). Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc. Quantitative comparisons against several prior methods demonstrate the superiority of our approach.