SEAN: Image Synthesis with Semantic Region-Adaptive Normalization

TL;DR

SEAN introduces a normalization technique for GANs that enables detailed control over the style of individual semantic regions in image synthesis, improving quality and interactivity.

Contribution

The paper presents SEAN, a novel normalization method that allows independent style control of semantic regions in GAN-based image synthesis, enhancing quality and editing capabilities.

Findings

Outperforms previous methods in FID and PSNR metrics

Enables interactive editing by changing segmentation masks or styles

Supports style interpolation between reference images

Abstract

We propose semantic region-adaptive normalization (SEAN), a simple but effective building block for Generative Adversarial Networks conditioned on segmentation masks that describe the semantic regions in the desired output image. Using SEAN normalization, we can build a network architecture that can control the style of each semantic region individually, e.g., we can specify one style reference image per region. SEAN is better suited to encode, transfer, and synthesize style than the best previous method in terms of reconstruction quality, variability, and visual quality. We evaluate SEAN on multiple datasets and report better quantitative metrics (e.g. FID, PSNR) than the current state of the art. SEAN also pushes the frontier of interactive image editing. We can interactively edit images by changing segmentation masks or the style for any given region. We can also interpolate styles from two reference images per region.