Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
read the original abstract
Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.
This paper has not been read by Pith yet.
Forward citations
Cited by 7 Pith papers
-
AGAN: Towards Automated Design of Generative Adversarial Networks
AGAN is the first neural architecture search method for GANs that discovers architectures outperforming state-of-the-art on CIFAR-10 unsupervised image generation and competitive on supervised tasks.
-
CKM Beyond Channel Gain: Spatial Correlation Map Construction with Deep Learning
A deep learning model decomposes and reconstructs spatial correlation maps from sparse samples using attention and multi-scale fusion, reporting cosine similarity above 0.8 on the CKMImageNet dataset.
-
Analyzing the Variety Loss in the Context of Probabilistic Trajectory Prediction
Proof and experiments show the MoN loss approximates the square root of the ground truth PDF in probabilistic trajectory prediction instead of the PDF.
-
Progressive Growing of GANs for Improved Quality, Stability, and Variation
Progressive growing stabilizes GAN training to produce high-resolution images of unprecedented quality and achieves a record unsupervised inception score of 8.80 on CIFAR10.
-
Deep Learning for MRI Slice Interpolation: The Critical Role of Problem Formulation
Reformulating the input to adjacent slices for deep learning MRI interpolation yields 58% SSIM gains and 10.1% improvement over linear baseline, with problem formulation outweighing architecture choice.
-
Image Enhancement by Recurrently-trained Super-resolution Network
Recurrent training of a super-resolution network by iteratively refining target images leads to improved enhancement quality.
-
Abnormal Colon Polyp Image Synthesis Using Conditional Adversarial Networks for Improved Detection Performance
A cGAN method with edge-filtered combined inputs generates synthetic polyp images from normal colonoscopy frames to augment training data and improve detection performance.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.