Coupled Generative Adversarial Networks
This addresses the challenge of data scarcity in multi-domain image analysis for computer vision researchers, offering a novel approach that reduces the need for paired training data.
The paper tackles the problem of learning joint distributions of multi-domain images without requiring corresponding image tuples in training, achieving this by using a weight-sharing constraint in coupled generative adversarial networks (CoGAN). It successfully applies CoGAN to tasks like color-depth image joint distribution and face attribute joint distribution, demonstrating applications in domain adaptation and image transformation.
We propose coupled generative adversarial network (CoGAN) for learning a joint distribution of multi-domain images. In contrast to the existing approaches, which require tuples of corresponding images in different domains in the training set, CoGAN can learn a joint distribution without any tuple of corresponding images. It can learn a joint distribution with just samples drawn from the marginal distributions. This is achieved by enforcing a weight-sharing constraint that limits the network capacity and favors a joint distribution solution over a product of marginal distributions one. We apply CoGAN to several joint distribution learning tasks, including learning a joint distribution of color and depth images, and learning a joint distribution of face images with different attributes. For each task it successfully learns the joint distribution without any tuple of corresponding images. We also demonstrate its applications to domain adaptation and image transformation.