Abstract:We propose a novel ECGAN for the challenging semantic image synthesis task. Although considerable improvements have been achieved by the community in the recent period, the quality of synthesized images is far from satisfactory due to three largely unresolved challenges. 1) The semantic labels do not provide detailed structural information, making it challenging to synthesize local details and structures; 2) The widely adopted CNN operations such as convolution, down-sampling, and normalization usually cause spatial resolution loss and thus cannot fully preserve the original semantic information, leading to semantically inconsistent results (e.g., missing small objects); 3) Existing semantic image synthesis methods focus on modeling 'local' semantic information from a single input semantic layout. However, they ignore 'global' semantic information of multiple input semantic layouts, i.e., semantic cross-relations between pixels across different input layouts. To tackle 1), we propose to use the edge as an intermediate representation which is further adopted to guide image generation via a proposed attention guided edge transfer module. To tackle 2), we design an effective module to selectively highlight class-dependent feature maps according to the original semantic layout to preserve the semantic information. To tackle 3), inspired by current methods in contrastive learning, we propose a novel contrastive learning method, which aims to enforce pixel embeddings belonging to the same semantic class to generate more similar image content than those from different classes. We further propose a novel multi-scale contrastive learning method that aims to push same-class features from different scales closer together being able to capture more semantic relations by explicitly exploring the structures of labeled pixels from multiple input semantic layouts from different scales.

BSD-GAN: Branched Generative Adversarial Network for Scale-Disentangled Representation Learning and Image Synthesis.

Branched Generative Adversarial Networks for Multi-Scale Image Manifold Learning.

SpatialGAN: Progressive Image Generation Based on Spatial Recursive Adversarial Expansion

Structured Generative Adversarial Networks.

Two Birds with One Stone: Iteratively Learn Facial Attributes with GANs.

A New Distributed Method for Training Generative Adversarial Networks

Densely Stacked Generative Adversarial Networks

Two Birds with One Stone: Transforming and Generating Facial Images with Iterative GAN

Large Scale GAN Training for High Fidelity Natural Image Synthesis

A Novel Generator with Auxiliary Branch for Improving GAN Performance

Boundary-Seeking Generative Adversarial Networks

Dual Attention GANs for Semantic Image Synthesis

ParaGAN: A Scalable Distributed Training Framework for Generative Adversarial Networks

GAN Dissection: Visualizing and Understanding Generative Adversarial Networks

Edge Guided GANs with Multi-Scale Contrastive Learning for Semantic Image Synthesis

JGAN: A Joint Formulation of GAN for Synthesizing Images and Labels

SP-GAN: Self-Growing and Pruning Generative Adversarial Networks

Stacked Siamese Generative Adversarial Nets: A Novel Way to Enlarge Image Dataset

Bayesian Conditional Generative Adverserial Networks

CVAE-GAN: Fine-Grained Image Generation through Asymmetric Training