ReMix: Towards Image-to-Image Translation with Limited Data
This addresses the issue of data scarcity for researchers and practitioners in computer vision, though it is incremental as it builds on existing GAN models.
The paper tackles the problem of overfitting in GAN-based image-to-image translation with limited data by proposing ReMix, a data augmentation method that interpolates training samples at the feature level and uses a novel content loss, resulting in significant improvements in performance across various tasks.
Image-to-image (I2I) translation methods based on generative adversarial networks (GANs) typically suffer from overfitting when limited training data is available. In this work, we propose a data augmentation method (ReMix) to tackle this issue. We interpolate training samples at the feature level and propose a novel content loss based on the perceptual relations among samples. The generator learns to translate the in-between samples rather than memorizing the training set, and thereby forces the discriminator to generalize. The proposed approach effectively reduces the ambiguity of generation and renders content-preserving results. The ReMix method can be easily incorporated into existing GAN models with minor modifications. Experimental results on numerous tasks demonstrate that GAN models equipped with the ReMix method achieve significant improvements.