StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks

Although Generative Adversarial Networks (GANs) have shown remarkable success in various tasks, they still face challenges in generating high quality images. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) aiming at generating high-resolution photo-realistic images... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Text-to-Image Generation COCO StackGAN-v1 FID 74.05 # 4
Inception score 8.45 # 11
Text-to-Image Generation COCO StackGAN-v2 FID 81.59 # 5
Inception score 8.30 # 12
Text-to-Image Generation CUB StackGAN-v2 FID 15.30 # 1
Inception score 3.82 # 5
Text-to-Image Generation CUB StackGAN-v1 FID 51.89 # 2
Inception score 3.70 # 6
Image Generation LSUN Bedroom 256 x 256 StackGAN-v2 FID 35.61 # 8
Text-to-Image Generation Oxford 102 Flowers StackGAN-v1 FID 55.28 # 2
Inception score 3.20 # 2
Text-to-Image Generation Oxford 102 Flowers StackGAN-v2 FID 48.68 # 1
Inception score 3.26 # 1

Methods used in the Paper