How good is my GAN?

About

Generative adversarial networks (GANs) are one of the most popular methods for generating images today. While impressive results have been validated by visual inspection, a number of quantitative criteria have emerged only recently. We argue here that the existing ones are insufficient and need to be in adequation with the task at hand. In this paper we introduce two measures based on image classification---GAN-train and GAN-test, which approximate the recall (diversity) and precision (quality of the image) of GANs respectively. We evaluate a number of recent GAN approaches based on these two measures and demonstrate a clear difference in performance. Furthermore, we observe that the increasing difficulty of the dataset, from CIFAR10 over CIFAR100 to ImageNet, shows an inverse correlation with the quality of the GANs, as clearly evident from our measures.

Konstantin Shmelkov, Cordelia Schmid, Karteek Alahari · 2018

Related benchmarks

Task	Dataset	Result
Image Classification	ImageNet (val)	Accuracy—	300
Image Classification	Stanford Cars 10% subset	Top-1 Acc18.91	5
Image Classification	Stanford Cars 25% subset	Top-1 Accuracy46.56	5
Image Classification	Stanford Cars 50%	Top-1 Accuracy68.81	5
Image Classification	Stanford Cars 75%	Top-1 Accuracy80.67	5
Classification	CIFAR-10-C 1.0 (test)	Accuracy (Clean)84.11	4