Progressive Growing of GANs for Improved Quality, Stability, and Variation

We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, we add new layers that model increasingly fine details as training progresses... (read more)

PDF Abstract ICLR 2018 PDF ICLR 2018 Abstract

Datasets


Introduced in the Paper:

CelebA-HQ

Mentioned in the Paper:

CIFAR-10 LSUN FFHQ

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Generation CelebA-HQ 1024x1024 PGGAN FID 7.3 # 4
Image Generation CelebA-HQ 256x256 PGGAN FID 8.03 # 1
Image Generation CIFAR-10 PGGAN Inception score 8.8 # 15
Image Generation FFHQ PGGAN FID 8.04 # 7
Image Generation LSUN Bedroom 256 x 256 PGGAN FID 8.34 # 6
Image Generation LSUN Cat 256 x 256 PGGAN FID 37.52 # 3
Image Generation LSUN Churches 256 x 256 PGGAN FID 6.42 # 5