arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2104.00487cs.CVcs.AI
收藏

生成对抗网络中的线性语义

Linear Semantics in Generative Adversarial Networks

Columbia University(哥伦比亚大学);Panzhihua University(攀枝花大学)

Jianjin Xu, Changxi Zheng

摘要:

生成对抗网络(GANs)能够生成高质量的图像,但仍然难以明确指定合成图像的语义。在这项工作中,我们的目标是更好地理解GANs的语义表示,从而使语义控制在GAN的生成过程。有趣的是,我们发现一个训练有素的GAN以一种令人惊讶的简单方式在其内部特征映射中编码图像语义:特征映射的线性变换足以提取生成的图像语义。为了验证这种简单性,我们对各种GANs和数据集进行了广泛的实验;由于这种简单性,我们能够从少量(例如,8)标签图像。最后,利用我们的研究结果,我们提出了两种Few-Shot图像编辑方法,即语义条件采样和语义图像编辑。给定训练的GAN和少至八个语义注释,用户能够生成服从用户提供的语义布局的各种图像,并控制合成的图像语义。我们已经公开了代码。

英文摘要:

Generative Adversarial Networks (GANs) are able to generate high-quality images, but it remains difficult to explicitly specify the semantics of synthesized images. In this work, we aim to better understand the semantic representation of GANs, and thereby enable semantic control in GAN's generation process. Interestingly, we find that a well-trained GAN encodes image semantics in its internal feature maps in a surprisingly simple way: a linear transformation of feature maps suffices to extract the generated image semantics. To verify this simplicity, we conduct extensive experiments on various GANs and datasets; and thanks to this simplicity, we are able to learn a semantic segmentation model for a trained GAN from a small number (e.g., 8) of labeled images. Last but not least, leveraging our findings, we propose two few-shot image editing approaches, namely Semantic-Conditional Sampling and Semantic Image Editing. Given a trained GAN and as few as eight semantic annotations, the user is able to generate diverse images subject to a user-provided semantic layout, and control the synthesized image semantics. We have made the code publicly available.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑