Leveraging Local Domains for Image-to-Image Translation

TL;DR

This paper introduces a method that uses local domain knowledge and geometrical guidance to improve image-to-image translation, enabling realistic translations with minimal data and enhancing downstream task performance.

Contribution

It proposes leveraging human spatial domain knowledge and patch-based GANs to generate unseen domain images, improving translation quality with limited data.

Findings

Realistic translations achieved with few images

Significant improvement in proxy task performance

Effective transfer learning to unseen domains

Abstract

Image-to-image (i2i) networks struggle to capture local changes because they do not affect the global scene structure. For example, translating from highway scenes to offroad, i2i networks easily focus on global color features but ignore obvious traits for humans like the absence of lane markings. In this paper, we leverage human knowledge about spatial domain characteristics which we refer to as 'local domains' and demonstrate its benefit for image-to-image translation. Relying on a simple geometrical guidance, we train a patch-based GAN on few source data and hallucinate a new unseen domain which subsequently eases transfer learning to target. We experiment on three tasks ranging from unstructured environments to adverse weather. Our comprehensive evaluation setting shows we are able to generate realistic translations, with minimal priors, and training only on a few images. Furthermore, when trained on our translations images we show that all tested proxy tasks are significantly improved, without ever seeing target domain at training.