Guided Image-to-Image Translation with Bi-Directional Feature Transformation
This addresses the need for more effective conditioning in image translation tasks, though it appears incremental as it builds on existing uni-directional methods.
The paper tackles the problem of guided image-to-image translation by introducing a bi-directional feature transformation scheme to better utilize user-provided guidance constraints, showing it outperforms other conditioning methods and achieves results comparable to state-of-the-art approaches.
We address the problem of guided image-to-image translation where we translate an input image into another while respecting the constraints provided by an external, user-provided guidance image. Various conditioning methods for leveraging the given guidance image have been explored, including input concatenation , feature concatenation, and conditional affine transformation of feature activations. All these conditioning mechanisms, however, are uni-directional, i.e., no information flow from the input image back to the guidance. To better utilize the constraints of the guidance image, we present a bi-directional feature transformation (bFT) scheme. We show that our bFT scheme outperforms other conditioning schemes and has comparable results to state-of-the-art methods on different tasks.