Filters








1,836 Hits in 3.5 sec

Unsupervised Real Image Super-Resolution via Generative Variational AutoEncoder

Zhi-Song Liu, Wan-Chi Siu, Li-Wen Wang, Chu-Tak Li, Marie-Paule Cani, Yui-Lam Chan
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
We come up with a conditional variational autoencoder to encode the reference for dense feature vector which can then be transferred to the decoder for target image denoising.  ...  In this paper, we revisit the classic example based image super-resolution approaches and come up with a novel generative model for perceptual image super-resolution.  ...  This is also the first work on combining Variational AutoEncoder and Generative Adversarial Network for image super-resolution. 3.  ... 
doi:10.1109/cvprw50498.2020.00229 dblp:conf/cvpr/LiuSWLCC20 fatcat:pefpfnflxrcw5jutze5iphey4u

Unsupervised Real Image Super-Resolution via Generative Variational AutoEncoder [article]

Zhi-Song Liu, Wan-Chi Siu, Li-Wen Wang, Chu-Tak Li, Marie-Paule Cani, Yui-Lam Chan
2020 arXiv   pre-print
We come up with a conditional variational autoencoder to encode the reference for dense feature vector which can then be transferred to the decoder for target image denoising.  ...  In this paper, we revisit the classic example based image super-resolution approaches and come up with a novel generative model for perceptual image super-resolution.  ...  This is also the first work on combining Variational AutoEncoder and Generative Adversarial Network for image super-resolution. 3.  ... 
arXiv:2004.12811v1 fatcat:iwrvcyfg4jamjijp3qv2j2eoi4

Reference based Face Super-resolution

Zhi-Song Liu, Wan-Chi Siu, Yui-Lam Chan
2019 IEEE Access  
We propose a novel Conditional Variational AutoEncoder model for this Reference based Face Super-Resolution (RefSR-VAE).  ...  We create a benchmark dataset on reference based face super-resolution (RefSR-Face) for general research use, which contains reference images paired with low-resolution images of various pose, emotions  ...  In this paper, we propose a novel Conditional Variational AutoEncoder model for Reference based Face Super-Resolution (RefSR-VAE).  ... 
doi:10.1109/access.2019.2934078 fatcat:qw6mh56ysvfpjbvno4hxn7sgwq

Unsupervised Super-Resolution: Creating High-Resolution Medical Images from Low-Resolution Anisotropic Examples [article]

Jörg Sander, Bob D. de Vos, Ivana Išgum
2020 arXiv   pre-print
To address this issue, we propose a learning-based super-resolution approach that can be trained using solely anisotropic images, i.e. without high-resolution ground truth data.  ...  The method exploits the latent space, generated by autoencoders trained on anisotropic images, to increase spatial resolution in low-resolution images.  ...  ACKNOWLEDGMENTS This study was performed within the DLMedIA program (P15-26) funded by Dutch Technology Foundation with participation of PIE Medical Imaging.  ... 
arXiv:2010.13172v1 fatcat:uqtgld3zdzht3ia4ql3rv6yvma

Image Super-Resolution With Deep Variational Autoencoders [article]

Darius Chira, Ilian Haralampiev, Ole Winther, Andrea Dittadi, Valentin Liévin
2022 arXiv   pre-print
Models based on Variational Autoencoders (VAEs) have often been criticized for their feeble generative performance, but with new advancements such as VDVAE (very deep VAE), there is now strong evidence  ...  Image super-resolution (SR) techniques are used to generate a high-resolution image from a low-resolution image.  ...  Conclusions In this paper, we propose VDVAE-SR, a Very Deep Variational Autoencoder (VDVAE) adapted for the task of image super-resolution (SR).  ... 
arXiv:2203.09445v1 fatcat:ymed7hhhenfkbaxhi3vom3qyle

Autoencoding Low-Resolution MRI for Semantically Smooth Interpolation of Anisotropic MRI [article]

Jörg Sander, Bob D. de Vos, Ivana Išgum
2022 arXiv   pre-print
Recently, better performing deep-learning based super-resolution methods have been introduced.  ...  High-resolution medical images are beneficial for analysis but their acquisition may not always be feasible.  ...  ACKNOWLEDGMENT This study was performed within the DLMedIA program (P15-26) funded by Dutch Technology Foundation with participation of Pie Medical Imaging.  ... 
arXiv:2202.09258v1 fatcat:ctyoztnoprg7vb3hgqwaljt6nq

Image Restoration using Autoencoding Priors [article]

Siavash Arjomand Bigdeli, Matthias Zwicker
2017 arXiv   pre-print
We demonstrate state of the art results for non-blind deconvolution and super-resolution using the same autoencoding prior.  ...  A key advantage of our approach is that we do not need to train separate networks for different image restoration tasks, such as non-blind deconvolution with different kernels, or super-resolution at different  ...  Classical techniques include priors based on edge statistics, total variation, sparse representations, or patch-based priors.  ... 
arXiv:1703.09964v1 fatcat:7f4tpcm6jvforcxufntm3c5lea

Model Inspired Autoencoder for Unsupervised Hyperspectral Image Super-Resolution [article]

Jianjun Liu, Zebin Wu, Liang Xiao, Xiao-Jun Wu
2021 arXiv   pre-print
This paper focuses on hyperspectral image (HSI) super-resolution that aims to fuse a low-spatial-resolution HSI and a high-spatial-resolution multispectral image to form a high-spatial-resolution HSI (  ...  Inspired by the specific properties of model, we make the first attempt to design a model inspired deep network for HSI super-resolution in an unsupervised manner.  ...  They would like to thank NCALM and the Hyperspectral Image Analysis Laboratory at UH for providing the UH datasets, and the Image Analysis and Data Fusion Technical Committee of the IEEE GRSS for supporting  ... 
arXiv:2110.11591v1 fatcat:6rxoim4qv5g3vejajliv3fqvpa

Bandwidth Extension on Raw Audio via Generative Adversarial Networks [article]

Sung Kim, Visvesh Sathe
2019 arXiv   pre-print
In this work we explore a GAN-based method for audio processing, and develop a convolutional neural network architecture to perform audio super-resolution.  ...  Neural network-based methods have recently demonstrated state-of-the-art results on image synthesis and super-resolution tasks, in particular by using variants of generative adversarial networks (GANs)  ...  Stacked autoencoders [43] and variational autoencoders [21, 37] have been used for denoising, image generation, and music synthesis [34] .  ... 
arXiv:1903.09027v1 fatcat:sbpiyc5kjjc3zj54l2funla6pu

Super-Resolving Very Low-Resolution Face Images with Supplementary Attributes

Xin Yu, Basura Fernando, Richard Hartley, Fatih Porikli
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
Given a tiny face image, existing face hallucination methods aim at super-resolving its high-resolution (HR) counterpart by learning a mapping from an exemplar dataset.  ...  We demonstrate that supplementing residual images or feature maps with additional facial attribute information can significantly reduce the ambiguity in face super-resolution.  ...  Aimed at addressing pose variations, part based methods super-resolve individual facial regions separately.  ... 
doi:10.1109/cvpr.2018.00101 dblp:conf/cvpr/YuFHP18 fatcat:eevz23ankvc6varp32hqgk67pu

VEHiCLE: a Variationally Encoded Hi-C Loss Enhancement algorithm [article]

Max R Highsmith, Jianlin R Cheng
2020 bioRxiv   pre-print
VEHiCLE utilizes a variational autoencoder and adversarial training strategy to enhance contact maps, making them more viable for downstream analysis.  ...  Using a variational autoencoder VEHiCLE provides a user tunable, full, generative model for generating synthetic Hi-C data while also providing state-of-the-art results in enhancement of Hi-C data across  ...  The use of autoencoders for the task of Hi-C data super resolution was originally proposed in our preprint 10 for the task of denoising Hi-C data.  ... 
doi:10.1101/2020.12.07.413559 fatcat:j6wry3vugbbsdd5xlwbcmq6amm

VEHiCLE: a Variationally Encoded Hi-C Loss Enhancement algorithm for improving and generating Hi-C data

Max Highsmith, Jianlin Cheng
2021 Scientific Reports  
Using a deep variational autoencoder, VEHiCLE provides a user tunable, full generative model for generating synthetic Hi-C data while also providing state-of-the-art results in enhancement of Hi-C data  ...  VEHiCLE expands previous efforts at Hi-C super resolution by providing novel insight into the biologically meaningful and human interpretable feature extraction.  ...  The use of autoencoders for the task of Hi-C data super resolution was originally proposed in our preprint 10 for the task of denoising Hi-C data.  ... 
doi:10.1038/s41598-021-88115-9 pmid:33893353 fatcat:iabf3lhfhbe3xgqb2oxpsjmmna

Learning to generate images with perceptual similarity metrics

Jake Snell, Karl Ridgeway, Renjie Liao, Brett D. Roads, Michael C. Mozer, Richard S. Zemel
2017 2017 IEEE International Conference on Image Processing (ICIP)  
Finally, we demonstrate the superiority of perceptually-optimized networks for super-resolution imaging.  ...  For three different architectures, we collected human judgments of the quality of image reconstructions.  ...  Visual comparisons on super-resolution at a magnification factor of 4.  ... 
doi:10.1109/icip.2017.8297089 dblp:conf/icip/SnellRLRMZ17 fatcat:ykczprlhjbhzbax5btzs7vnjai

Model-based Reconstruction with Learning: From Unsupervised to Supervised and Beyond [article]

Zhishen Huang and Siqi Ye and Michael T. McCann and Saiprasad Ravishankar
2021 arXiv   pre-print
For example, sparsity or low-rankness based regularizers have been widely used for image reconstruction from limited data such as in compressed sensing.  ...  Learning-based approaches for image reconstruction have garnered much attention in recent years and have shown promise across biomedical imaging applications.  ...  Autoencoders have been incorporated into MBIR frameworks to provide learning-based priors, but most of these works train autoencoders in a supervised manner based on paired noisy and reference images  ... 
arXiv:2103.14528v1 fatcat:kxzugqnnijdwfn62jwrl45zmge

Learning to Generate Images with Perceptual Similarity Metrics [article]

Jake Snell, Karl Ridgeway, Renjie Liao, Brett D. Roads, Michael C. Mozer, Richard S. Zemel
2017 arXiv   pre-print
Finally, we demonstrate the superiority of perceptually-optimized networks for super-resolution imaging.  ...  For three different architectures, we collected human judgments of the quality of image reconstructions.  ...  Image Super-Resolution Details For the super-resolution experiments, all input images are converted from RGB to YCbCr color space and only Y channel is used for training and testing.  ... 
arXiv:1511.06409v3 fatcat:ig2kqqvucjcihbogroo5djy3bi
« Previous Showing results 1 — 15 out of 1,836 results