Leveraging Local Domains for Image-to-Image Translation
This work addresses a specific bottleneck in image-to-image translation for applications like unstructured environments and adverse weather, offering an incremental improvement over existing methods.
The paper tackles the problem of image-to-image translation networks struggling with local changes by leveraging human knowledge of spatial domain characteristics, termed 'local domains', and demonstrates that this approach generates realistic translations with minimal priors and few training images, improving proxy tasks without target domain data.
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.