Learning to Predict Layout-to-image Conditional Convolutions for Semantic Image Synthesis

TL;DR

This paper introduces a novel method for semantic image synthesis that predicts convolutional kernels conditioned on semantic layouts, improving the quality and semantic alignment of generated images.

Contribution

It proposes a kernel prediction approach conditioned on semantic maps and a feature pyramid discriminator, advancing the state-of-the-art in photorealistic image synthesis from semantic layouts.

Findings

Achieves state-of-the-art quantitative metrics.

Produces more detailed and semantically aligned images.

Outperforms previous methods on multiple datasets.

Abstract

Semantic image synthesis aims at generating photorealistic images from semantic layouts. Previous approaches with conditional generative adversarial networks (GAN) show state-of-the-art performance on this task, which either feed the semantic label maps as inputs to the generator, or use them to modulate the activations in normalization layers via affine transformations. We argue that convolutional kernels in the generator should be aware of the distinct semantic labels at different locations when generating images. In order to better exploit the semantic layout for the image generator, we propose to predict convolutional kernels conditioned on the semantic label map to generate the intermediate feature maps from the noise maps and eventually generate the images. Moreover, we propose a feature pyramid semantics-embedding discriminator, which is more effective in enhancing fine details and semantic alignments between the generated images and the input semantic layouts than previous multi-scale discriminators. We achieve state-of-the-art results on both quantitative metrics and subjective evaluation on various semantic segmentation datasets, demonstrating the effectiveness of our approach.