[2102.06108] SWAGAN: A Style-based Wavelet-driven Generative Model

In recent years, considerable progress has been made in the visual quality of Generative Adversarial Networks (GANs). Even so, these networks still suffer from degradation in quality for high-frequency content, stemming from a spectrally biased architecture, and similarly unfavorable loss functions. To address this issue, we present a novel general-purpose Style and WAvelet based GAN (SWAGAN) that implements progressive generation in the frequency domain. SWAGAN incorporates wavelets throughout its generator and discriminator architectures, enforcing a frequency-aware latent representation at every step of the way. This approach yields enhancements in the visual quality of the generated images, and considerably increases computational performance. We demonstrate the advantage of our method by integrating it into the SyleGAN2 framework, and verifying that content generation in the wavelet domain leads to higher quality images with more realistic high-frequency content. Furthermore, we v

5 mentions: @ak92501@mosko_mule@AkiraTOSEI@AkiraTOSEI@arxiv_cs_cv_pr
Keywords: gan
Date: 2021/02/19 17:22

Referring Tweets

@mosko_mule SWAGAN: A Style-based Wavelet-driven Generative Model t.co/w4rlhal32I StyleGAN2のアーキテクチャをwavelet係数の予測に置き換えたことでStyleGAN2より実時間、イテレーション数ともに高速に良いFIDに達する。補完もより自然な気がする。 t.co/a5M5eLzaE2
@AkiraTOSEI t.co/EV8JR7hAVi StyleGANにおいて、RGB画像ではなくwavelet情報をProgressive-Growingさせることで高解像度画像を生成させる研究。1/4程度の計算資源だけでRGBベースの手法と同等の品質にまで到達させることができる。 t.co/icAGvNEgqs
@AkiraTOSEI t.co/EV8JR7003K In StyleGAN, this study generates high-resolution images by Progressive-Growing wavelets instead of RGB images, and it can achieve the same quality as RGB-based methods with only 1/4 of the computational resources. t.co/LlUVMlsI5g
@ak92501 SWAGAN: A Style-based WAvelet-driven Generative Model pdf: t.co/pp9p90Sc5M abs: t.co/PjaI9xRYEn t.co/cY8sHU0g3U

Related Entries

Read more [2008.00951] Encoding in Style: a StyleGAN Encoder for Image-to-Image Translationopen searchopen nav...
0 users, 3 mentions 2020/08/11 14:21
Read more [2009.13829] TinyGAN: Distilling BigGAN for Conditional Image Generationopen searchopen navigation m...
0 users, 4 mentions 2020/10/07 11:21
Read more [2011.15126] One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencingopen searchopen ...
0 users, 4 mentions 2020/12/01 03:51
Read more [2102.05095] Is Space-Time Attention All You Need for Video Understanding?
0 users, 6 mentions 2021/02/13 11:21
Read more [2102.06529] Improving Object Detection in Art Images Using Only Style Transfer
0 users, 3 mentions 2021/02/19 11:21