Leveraging in-domain supervision for unsupervised image-to-image translation tasks via multi-stream generators
This work addresses the challenge of improving translation quality in UI2I scenarios where supervision is scarce, particularly for complex scenes like urban environments, though it appears incremental by building on existing UI2I methods.
The paper tackles the problem of unsupervised image-to-image translation (UI2I) by incorporating in-domain prior knowledge like semantic segmentation to improve translation quality, demonstrating superior results in converting day images to night ones on urban data and showing that augmented images enhance downstream detection tasks.
Supervision for image-to-image translation (I2I) tasks is hard to come by, but bears significant effect on the resulting quality. In this paper, we observe that for many Unsupervised I2I (UI2I) scenarios, one domain is more familiar than the other, and offers in-domain prior knowledge, such as semantic segmentation. We argue that for complex scenes, figuring out the semantic structure of the domain is hard, especially with no supervision, but is an important part of a successful I2I operation. We hence introduce two techniques to incorporate this invaluable in-domain prior knowledge for the benefit of translation quality: through a novel Multi-Stream generator architecture, and through a semantic segmentation-based regularization loss term. In essence, we propose splitting the input data according to semantic masks, explicitly guiding the network to different behavior for the different regions of the image. In addition, we propose training a semantic segmentation network along with the translation task, and to leverage this output as a loss term that improves robustness. We validate our approach on urban data, demonstrating superior quality in the challenging UI2I tasks of converting day images to night ones. In addition, we also demonstrate how reinforcing the target dataset with our augmented images improves the training of downstream tasks such as the classical detection one.