Semantic Scholar Open Access 2018 12630 sitasi

A Style-Based Generator Architecture for Generative Adversarial Networks

Tero Karras S. Laine Timo Aila

Lihat Sumber DOI

Abstrak

We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis. The new generator improves the state-of-the-art in terms of traditional distribution quality metrics, leads to demonstrably better interpolation properties, and also better disentangles the latent factors of variation. To quantify interpolation quality and disentanglement, we propose two new, automated methods that are applicable to any generator architecture. Finally, we introduce a new, highly varied and high-quality dataset of human faces.

Topik & Kata Kunci

Computer Science Mathematics

Penulis (3)

Tero Karras

S. Laine

Timo Aila

Format Sitasi

APA MLA BibTeX

Karras, T., Laine, S., Aila, T. (2018). A Style-Based Generator Architecture for Generative Adversarial Networks. https://doi.org/10.1109/CVPR.2019.00453

Akses Cepat

Lihat di Sumber doi.org/10.1109/CVPR.2019.00453

Informasi Jurnal

Tahun Terbit: 2018
Bahasa: en
Total Sitasi: 12630×
Sumber Database: Semantic Scholar
DOI: 10.1109/CVPR.2019.00453
Akses: Open Access ✓