Large Scale GAN Training for High Fidelity Natural Image Synthesis

ICLR 2019 Andrew BrockJeff DonahueKaren Simonyan

Despite recent progress in generative image modeling, successfully generating high-resolution, diverse samples from complex datasets such as ImageNet remains an elusive goal. To this end, we train Generative Adversarial Networks at the largest scale yet attempted, and study the instabilities specific to such scale... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Conditional Image Generation ImageNet 128x128 BigGAN-deep FID 7.4 # 3
Inception score 166.5 # 1
Image Generation ImageNet 128x128 BigGAN FID 8.7 # 3
IS 98.8 # 2
Conditional Image Generation ImageNet 128x128 BigGAN FID 9.6 # 5
Inception score 166.3 # 2