Quality Aware Generative Adversarial Networks

By Parimala Kancharla, Sumohana S. Channappayya. 2019

29 citations recorded by the source corpus.

Generative Adversarial Networks (GANs) have become a very popular tool for im- plicitly learning high-dimensional probability distributions. Several improvements have been made to the original GAN formulation to address some of its shortcom- ings like mode collapse, convergence issues, entanglement, poor visual quality etc. While a significant effort has been directed towards improving the visual quality of images generated by GANs, it is rather surprising that objective image quality metrics have neither been employed as cost functions nor as regularizers in GAN objective functions. In this work, we show how a distance metric that is a variant of the Structural SIMilarity (SSIM) index (a popular full-reference image quality assessment algorithm), and a novel quality aware discriminator gradient penalty function that is inspired by the Natural Image Quality Evaluator (NIQE, a popular no-reference image quality assessment algorithm) can each be used as excellent regularizers for GAN objective functions. Specifically, we demonstrate state-of- the-art performance using the Wasserstein GAN gradient penalty (WGAN-GP) framework over CIFAR-10, STL10 and CelebA datasets.

Subfields: Computer Vision and Pattern Recognition, Generative Models

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Fang Wen
2. Dong Chen
3. Jianmin Bao
4. Shuyang Gu
5. Xiaoou Tang
6. Xiaogang Wang
7. Luc Van Gool
8. Il-Chul Moon

In Generative Models

1. Dong Chen
2. Fang Wen
3. Il-Chul Moon
4. Wanmo Kang
5. Shuyang Gu
6. Seungjae Shin
7. Kyungwoo Song
8. Dongjun Kim

Citations in the corpus

• Score Matching Model for Unbounded Data Score (2021, methodology): “It is common practice [44] in the GAN community to use PSNR and SSIM [45] for the image-wise quality assessment, but those metrics require the noise-free reference images to compare the structural similarity with, in which our setting does not permit such clean images.”
• PriorGAN: Real Data Prior for Generative Adversarial Nets (2020, background): “More recent works [7, 21, 5] design task-specific loss functions for GANs.”
• NCP-VAE: Variational Autoencoders with Noise Contrastive Priors (2020, unknown): “Note how reconstruction FID is much lower than our FID, indicating that our (Vahdat & Kautz, 2020) 13.48 RAE (Ghosh et al., 2020) 40.95 2s-VAE (Dai & Wipf, 2018) 44.4 WAE (Tolstikhin et al., 2018) 35 Perceptial AE(Zhang et al., 2020) 13.8 Latent EBM (Pang et al., 2020) 37.87 COCO-GAN (Lin et al., 20”