Filters








29,408 Hits in 5.9 sec

Training End-to-end Single Image Generators without GANs [article]

Yael Vinker and Nir Zabari and Yedid Hoshen
2020 arXiv   pre-print
We present AugurOne, a novel approach for training single image generative models.  ...  We experimentally evaluate our method and show that it obtains compelling novel animations of single-image, as well as, state-of-the-art performance on conditional generation tasks e.g. paint-to-image  ...  This lead to the development of a novel non-adversarial approach for single image generative modeling.  ... 
arXiv:2004.06014v1 fatcat:dwmb5izo25hqfh4qg2od37qjp4

D-SRGAN: DEM Super-Resolution with Generative Adversarial Networks [article]

Bekir Z Demiray, Muhammed Sit, Ibrahim Demir
2020 arXiv   pre-print
Numerous new methods have been proposed such as Generative Adversarial Networks (GANs) to create intelligent models that correct and augment large-scale datasets.  ...  With recent development in Graphical Processing Units (GPU) and novel algorithms, deep learning techniques have become attractive to researchers for their performance in learning features from high-resolution  ...  CONCLUSIONS In this study, a generative adversarial network, D-SRGAN, is proposed. D-SRGAN aims to convert low-resolution DEMs into high-resolution ones without needing additional information.  ... 
arXiv:2004.04788v2 fatcat:drlchnpmzzgglbwaen2mtozcmm

Text to Image Translation using GAN with NLP and Computer Vision

2022 TAGA Journal  
To improve the generated images' variety and regulate the conditional-GAN training, we introduce a novel Conditioning Augmentation technique.  ...  This paper proposes Stacked Generative Adversarial Networks (StackGAN) to generate 256 x 256 photo-realistic images conditioned on text descriptions.  ...  Stacked Generative Adversarial Networks to generate high-resolution images with photo-realistic details, we propose a simple yet effective Stacked Generative Adversarial Networks.  ... 
doi:10.37896/pd91.4/91449 fatcat:cozyntnmkngkrn7ys3oaxhrwdq

A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution

Rafael E. Rivadeneira, Angel D. Sappa, Boris X. Vintimilla, Riad Hammoud
2022 Sensors  
This paper presents a transfer domain strategy to tackle the limitations of low-resolution thermal sensors and generate higher-resolution images of reasonable quality.  ...  The proposed technique employs a CycleGAN architecture and uses a ResNet as an encoder in the generator along with an attention module and a novel loss function.  ...  a high variability needed to generalize the training.  ... 
doi:10.3390/s22062254 pmid:35336426 pmcid:PMC8953585 fatcat:l4tq3nh52repfbgx5dxfh4mkgq

Generalized Real-World Super-Resolution through Adversarial Robustness [article]

Angela Castillo, María Escobar, Juan C. Pérez, Andrés Romero, Radu Timofte, Luc Van Gool, Pablo Arbeláez
2021 arXiv   pre-print
In contrast to the traditional proposal, we present Robust Super-Resolution (RSR), a method that leverages the generalization capability of adversarial attacks to tackle real-world SR.  ...  Instead of learning a dataset-specific degradation, we employ adversarial attacks to create difficult examples that target the model's weaknesses.  ...  We present RSR, a novel SR method that leverages robust adversarial examples to create photo-realistic HR images regardless of the LR input noise.  ... 
arXiv:2108.11505v1 fatcat:irsrdqyzabc6zli3nk4a4cfi2m

D-SRGAN: DEM Super-Resolution with Generative Adversarial Networks

Bekir Z. Demiray, Muhammed Sit, Ibrahim Demir
2021 SN Computer Science  
In this paper, a GAN based model (D-SRGAN), inspired by single image super-resolution methods, is developed and evaluated to increase the resolution of DEMs.  ...  The goal of this study is to develop a machine learning model that increases the spatial resolution of DEM without additional information.  ...  Main contributions of this paper are: (a) proposing a new generative adversarial network (D-SRGAN) based approach for increasing the resolution of 50-feet DEMs to the resolution of 3-feet DEMs, (b) showing  ... 
doi:10.1007/s42979-020-00442-2 fatcat:v55ng2neyrcoriveo25skhnqfe

High-resolution Deep Convolutional Generative Adversarial Networks [article]

J. D. Curtó and I. C. Zarza and Fernando de la Torre and Irwin King and Michael R. Lyu
2020 arXiv   pre-print
Generative Adversarial Networks (GANs) [Goodfellow et al. 2014] convergence in a high-resolution setting with a computational constrain of GPU memory capacity has been beset with difficulty due to the  ...  In order to boost network convergence of DCGAN (Deep Convolutional Generative Adversarial Networks) [Radford et al. 2016] and achieve good-looking high-resolution results we propose a new layered network  ...  High-resolution Deep Convolutional Generative Adversarial Networks. • 7 Besides, to illustrate how fundamental our approach is, we enlarge To exemplify that the model is generating  ... 
arXiv:1711.06491v18 fatcat:yfvv3mge3vgbrffy3gts3ae45a

High Dimensional Spaces, Deep Learning and Adversarial Examples [article]

Simant Dube
2018 arXiv   pre-print
We point out mistake in an argument presented in prior published literature, and we present a more rigorous, general and correct mathematical result to explain adversarial examples in terms of topology  ...  In this paper, we analyze deep learning from a mathematical point of view and derive several novel results. The results are based on intriguing mathematical properties of high dimensional spaces.  ...  Though high dimensions pose a challenge, we can solve these challenges using novel ways of exploiting characteristics of natural images which will eliminate adversarial examples and overcome the current  ... 
arXiv:1801.00634v5 fatcat:asephmeud5grfjyl5ql53g6zma

Adversarial Defense by Stratified Convolutional Sparse Coding [article]

Bo Sun, Nian-hsuan Tsai, Fangchen Liu, Ronald Yu, Hao Su
2019 arXiv   pre-print
We propose an adversarial defense method that achieves state-of-the-art performance among attack-agnostic adversarial defense methods while also maintaining robustness to input resolution, scale of adversarial  ...  We introduce a novel Sparse Transformation Layer (STL) in between the input image and the first layer of the neural network to efficiently project images into our quasi-natural image space.  ...  [28] generate strongly transferable adversarial examples with an ensemble-based approach.  ... 
arXiv:1812.00037v2 fatcat:lsvxavf3jnh3xkno2pu67zcttm

Ad2Attack: Adaptive Adversarial Attack on Real-Time UAV Tracking [article]

Changhong Fu, Sihang Li, Xinnan Yuan, Junjie Ye, Ziang Cao, Fangqiang Ding
2022 arXiv   pre-print
Specifically, adversarial examples are generated online during the resampling of the search patch image, which leads trackers to lose the target in the following frames.  ...  Therefore, to help increase awareness of the potential risk and the robustness of UAV tracking, this work proposes a novel adaptive adversarial attack approach, i.e., Ad^2Attack, against UAV object tracking  ...  Subsequently, the LR images are resampled using a carefully built super-resolution upsampling (SrU) network to generate adversarial examples against UAV tracking.  ... 
arXiv:2203.01516v1 fatcat:4dichgbjlbf7te4ml43iaslt3q

Fine-grained Synthesis of Unrestricted Adversarial Examples [article]

Omid Poursaeed, Tianxing Jiang, Yordanos Goshu, Harry Yang, Serge Belongie, Ser-Nam Lim
2020 arXiv   pre-print
We propose a novel approach for generating unrestricted adversarial examples by manipulating fine-grained aspects of image generation.  ...  We perform experiments on LSUN, CelebA-HQ and COCO-Stuff as high resolution datasets to validate efficacy of our proposed approach.  ...  High resolution versions of adversarial examples. From left to right: original, noise-based and style-based images.  ... 
arXiv:1911.09058v2 fatcat:idb75q4bqzhwhhbzt3webh5fv4

Adversarial Defense by Stratified Convolutional Sparse Coding

Bo Sun, Nian-Hsuan Tsai, Fangchen Liu, Ronald Yu, Hao Su
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
We propose an adversarial defense method that achieves state-of-the-art performance among attack-agnostic adversarial defense methods while also maintaining robustness to input resolution, scale of adversarial  ...  We introduce a novel Sparse Transformation Layer (STL) between the input image and the first layer of the neural network to efficiently project images into our quasi-natural image space.  ...  [27] generate strongly transferable adversarial examples with an ensemble-based approach.  ... 
doi:10.1109/cvpr.2019.01171 dblp:conf/cvpr/SunTLYS19 fatcat:lg6gyvlq7bdatdgzuqmdqqs5vy

VoLux-GAN: A Generative Model for 3D Face Synthesis with HDRI Relighting [article]

Feitong Tan, Sean Fanello, Abhimitra Meka, Sergio Orts-Escolano, Danhang Tang, Rohit Pandey, Jonathan Taylor, Ping Tan, Yinda Zhang
2022 arXiv   pre-print
We propose VoLux-GAN, a generative framework to synthesize 3D-aware faces with convincing relighting.  ...  Multiple experiments and comparisons with other generative frameworks show how our model is a step forward towards photorealistic relightable 3D generative models.  ...  We summarize the contributions of this paper: 1) We propose a novel approach to generate HDRI relightable 3D faces with a volumetric rendering framework. 2) Supervised adversary losses are leveraged to  ... 
arXiv:2201.04873v1 fatcat:hkxwjzghvzanhgtw7vardumrqe

Full-Body High-Resolution Anime Generation with Progressive Structure-Conditional Generative Adversarial Networks [chapter]

Koichi Hamada, Kentaro Tachibana, Tianqi Li, Hiroto Honda, Yusuke Uchida
2019 Lecture Notes in Computer Science  
Recent progress in generative adversarial networks with progressive training has made it possible to generate high-resolution images.  ...  We propose Progressive Structure-conditional Generative Adversarial Networks (PSGAN), a new framework that can generate fullbody and high-resolution character images based on structural information.  ...  Automatic Dataset Construction with Exact Pose Keypoints from Unity 3D Models We create a novel dataset containing full-body high-resolution anime character images and exact 2D pose keypoints using the  ... 
doi:10.1007/978-3-030-11015-4_8 fatcat:ctpgirgj4nfa3gdc25l4s5nrna

Comparison of Al-Based Approaches for Statistical Downscaling of Surface Wind Fields in the North Atlantic

Vadim Rezvov, Mikhail Krinitskiy, Alexander Gavrikov, Sergey Gulev
2021 Zenodo  
Deep learning methods are one of the typical examples of the machine learning approaches to complex nonlinear functions approximating.  ...  Accurate prediction of high-resolution near-surface winds has a wide variety of applications.  ...  pressure Methods The first downscaling method used in this work is cubic interpolation from a low-resolution to a high-resolution grid.  ... 
doi:10.5281/zenodo.5760066 fatcat:cuvebscgrnbahbijg5vmgu376e
« Previous Showing results 1 — 15 out of 29,408 results