The Spatially-Correlative Loss for Various Image Translation Tasks
arXiv:2104.00854
Abstract
We propose a novel spatially-correlative loss that is simple, efficient and yet effective for preserving scene structure consistency while supporting large appearance changes during unpaired image-to-image (I2I) translation. Previous methods attempt this by using pixel-level cycle-consistency or feature-level matching losses, but the domain-specific nature of these losses hinder translation across large domain gaps. To address this, we exploit the spatial patterns of self-similarity as a means of defining scene structure. Our spatially-correlative loss is geared towards only capturing spatial relationships within an image rather than domain appearance. We also introduce a new self-supervised learning method to explicitly learn spatially-correlative maps for each specific translation task. We show distinct improvement over baseline models in all three modes of unpaired I2I translation: single-modal, multi-modal, and even single-image translation. This new loss can easily be integrated into existing network architectures and thus allows wide applicability.
arXiv comment14 pages, 12 figures
References in corpus (10)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
- A Simple Framework for Contrastive Learning of Visual Representations
- GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
- Learning deep representations by mutual information estimation and maximization
- Data-Efficient Image Recognition with Contrastive Predictive Coding
- Which Training Methods for GANs do actually Converge?
- Learning to Discover Cross-Domain Relations with Generative Adversarial Networks
- Unsupervised Attention-guided Image to Image Translation
- DRIT++: Diverse Image-to-Image Translation via Disentangled Representations