Interpreting Class Conditional GANs with Channel Awareness

TL;DR

This paper investigates how class-conditional GANs utilize feature channels for different categories, introducing channel awareness to interpret and manipulate the generative process, with applications in editing, hybridization, segmentation, and performance evaluation.

Contribution

It introduces channel awareness as a novel method to interpret class-conditional GANs and demonstrates its effectiveness in understanding and manipulating generated images.

Findings

Only a subset of channels are responsible for specific categories.

Related categories share some common channels.

Channels can be used for image editing and hybridization.

Abstract

Understanding the mechanism of generative adversarial networks (GANs) helps us better use GANs for downstream applications. Existing efforts mainly target interpreting unconditional models, leaving it less explored how a conditional GAN learns to render images regarding various categories. This work fills in this gap by investigating how a class conditional generator unifies the synthesis of multiple classes. For this purpose, we dive into the widely used class-conditional batch normalization (CCBN), and observe that each feature channel is activated at varying degrees given different categorical embeddings. To describe such a phenomenon, we propose channel awareness, which quantitatively characterizes how a single channel contributes to the final synthesis. Extensive evaluations and analyses on the BigGAN model pre-trained on ImageNet reveal that only a subset of channels is primarily responsible for the generation of a particular category, similar categories (e.g., cat and dog) usually get related to some same channels, and some channels turn out to share information across all classes. For good measure, our algorithm enables several novel applications with conditional GANs. Concretely, we achieve (1) versatile image editing via simply altering a single channel and manage to (2) harmoniously hybridize two different classes. We further verify that the proposed channel awareness shows promising potential in (3) segmenting the synthesized image and (4) evaluating the category-wise synthesis performance.