Current browse context:
cs.CV
Change to browse by:
References & Citations
Computer Science > Computer Vision and Pattern Recognition
Title: MobileStyleGAN: A Lightweight Convolutional Neural Network for High-Fidelity Image Synthesis
(Submitted on 10 Apr 2021 (v1), last revised 9 Sep 2021 (this version, v2))
Abstract: In recent years, the use of Generative Adversarial Networks (GANs) has become very popular in generative image modeling. While style-based GAN architectures yield state-of-the-art results in high-fidelity image synthesis, computationally, they are highly complex. In our work, we focus on the performance optimization of style-based generative models. We analyze the most computationally hard parts of StyleGAN2, and propose changes in the generator network to make it possible to deploy style-based generative networks in the edge devices. We introduce MobileStyleGAN architecture, which has x3.5 fewer parameters and is x9.5 less computationally complex than StyleGAN2, while providing comparable quality.
Submission history
From: Sergei Belousov [view email][v1] Sat, 10 Apr 2021 13:46:49 GMT (3079kb,D)
[v2] Thu, 9 Sep 2021 21:03:58 GMT (3082kb,D)
Link back to: arXiv, form interface, contact.