30,521 Hits in 9.9 sec

Informative Sample Mining Network for Multi-Domain Image-to-Image Translation [article]

Jie Cao, Huaibo Huang, Yi Li, Ran He, Zhenan Sun
2020 arXiv   pre-print
To select informative samples, we dynamically estimate sample importance during the training of Generative Adversarial Networks, presenting Informative Sample Mining Network.  ...  The performance of multi-domain image-to-image translation has been significantly improved by recent progress in deep generative models.  ...  In this paper, we propose Informative sample mining network (INIT) to enhance training efficiency and improve performance in multi-domain I2I tasks.  ... 
arXiv:2001.01173v4 fatcat:og7pixtnxngw5lyfdywika42ru

Exploring Patch-wise Semantic Relation for Contrastive Learning in Image-to-Image Translation Tasks [article]

Chanyong Jung, Gihyun Kwon, Jong Chul Ye
2022 arXiv   pre-print
We verified our method for three tasks: single-modal and multi-modal image translations, and GAN compression task for image translation.  ...  To further improve the performance, we present a hard negative mining by exploiting the semantic relation.  ...  Multi-modal image translation For further evaluation, we apply our method to multimodal image translation model which is a framework for translating input to diverse outputs with multiple domains.  ... 
arXiv:2203.01532v1 fatcat:zbepagm2fngdjcdns7j4g4oeeu

Instance-wise Hard Negative Example Generation for Contrastive Learning in Unpaired Image-to-Image Translation [article]

Weilun Wang, Wengang Zhou, Jianmin Bao, Dong Chen, Houqiang Li
2021 arXiv   pre-print
To address this issue, we present instance-wise hard Negative Example Generation for Contrastive learning in Unpaired image-to-image Translation (NEGCUT).  ...  In this paper, we uncover that the negative examples play a critical role in the performance of contrastive learning for image translation.  ...  It was also supported by the GPU cluster built by MCC Lab of Information Science and Technology Institution, USTC.  ... 
arXiv:2108.04547v2 fatcat:n3vtsbceunbvzdrbxz2owmfrpe

Deep Learning for SAR-Optical Image Matching

Lloyd Haydn Hughes, Nina Merkle, Tatjana Burgmann, Stefan Auer, Michael Schmitt
2019 IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium  
Driven by the success of deep learning in conventional optical image matching, we have carried out extensive research with regard to deep matching for SAR-optical multi-sensor image pairs in the recent  ...  underlying loss function, and creation of artificial images by generative adversarial networks.  ...  Fig. 2 : 2 Fig. 2: Conditional GAN architecture: the generator network learns to translate images between domains, while the discriminator learns to distinguish generated and real image pairs.  ... 
doi:10.1109/igarss.2019.8898635 dblp:conf/igarss/HughesMBA019 fatcat:jtl3oq4gnfb5pahswozgs4sizq

Image-to-Image Translation: Methods and Applications [article]

Yingxue Pang, Jianxin Lin, Tao Qin, Zhibo Chen
2021 arXiv   pre-print
Image-to-image translation (I2I) aims to transfer images from a source domain to a target domain while preserving the content representations.  ...  I2I has drawn increasing attention and made tremendous progress in recent years because of its wide range of applications in many computer vision and image processing problems, such as image synthesis,  ...  [168] propose the informative sample mining network (INIT) to analyze the importance of sample selection and select the informative samples for multihop training. Wu et al.  ... 
arXiv:2101.08629v2 fatcat:i6pywjwnvnhp3i7cmgza2slnle

GD-StarGAN: Multi-domain image-to-image translation in garment design

Yangyun Shen, Runnan Huang, Wenkai Huang, Yuanquan Wang
2020 PLoS ONE  
One of the image-to-image translation models--StarGAN, has realized the function of multi-domain image-to-image translation by using only a single generator and a single discriminator.  ...  In the field of fashion design, designing garment image according to texture is actually changing the shape of texture image, and image-to-image translation based on Generative Adversarial Network (GAN  ...  L1 loss is for two domains and is not suitable for multi-domain image-to-image translation. Thus, the domain classification loss is adopted in this paper to achieve cyclic consistency.  ... 
doi:10.1371/journal.pone.0231719 pmid:32315361 fatcat:sgcalmpzcjhftcfgpllycevryu

MCMI: Multi-Cycle Image Translation with Mutual Information Constraints [article]

Xiang Xu, Megha Nawhal, Greg Mori, Manolis Savva
2020 arXiv   pre-print
We present a mutual information-based framework for unsupervised image-to-image translation.  ...  The proposed mutual information constraints can improve cross-domain mappings by optimizing out translation functions that fail to satisfy the Markov property during image translations.  ...  Introduction Image-to-image (I2I) translation has gained prominence in recent years. The goal for I2I translation is to map an image from one domain to a corresponding image in another domain.  ... 
arXiv:2007.02919v1 fatcat:3xbmuymkhjf33nnxmeggysw44u

Biphasic Learning of GANs for High-Resolution Image-to-Image Translation [article]

Jie Cao, Huaibo Huang, Yi Li, Jingtuo Liu, Ran He, Zhenan Sun
2019 arXiv   pre-print
In this work, we present a novel training framework for GANs, namely biphasic learning, to achieve image-to-image translation in multiple visual domains at 1024^2 resolution.  ...  and sample quality when applied to the high-resolution situation.  ...  loss (denoted as "w/o mutual Info loss"), a network that maximizes mutual information as described in MINE [1] (denoted as "w/ MINE loss"), and a network that replaces mutual information loss with the  ... 
arXiv:1904.06624v1 fatcat:in6vb4wnczhhtbczsalm4rvwuy

Unsupervised Domain Adaptation of Object Detectors: A Survey [article]

Poojan Oza, Vishwanath A. Sindagi, Vibashan VS, Vishal M. Patel
2021 arXiv   pre-print
Due to this, model performance drops drastically when evaluated on label-scarce datasets having visually distinct images, termed as domain adaptation problem.  ...  There is a plethora of works to adapt classification and segmentation models to label-scarce target datasets through unsupervised domain adaptation.  ...  [121] Adversarial feature learning Image-to-image translation Domain randomization Roychowdhury et al. [61] Khodabandeh et al. [62] Kim et al. [97] D'Innocente et al.  ... 
arXiv:2105.13502v2 fatcat:ozzbbvoflfdvjdewjnjmfajlpa

Leveraging Virtual and Real Person for Unsupervised Person Re-identification [article]

Fengxiang Yang, Zhun Zhong, Zhiming Luo, Sheng Lian, Shaozi Li
2018 arXiv   pre-print
To address this problem, this study introduces a novel approach for unsupervised person re-ID by leveraging virtual and real data.  ...  For training of deep re-ID model, we divide it into three steps: 1) pre-training a coarse re-ID model by using virtual data; 2) collaborative filtering based positive pair mining from the real data; and  ...  These two models can only transfer images from one domain to another and may not be flexible enough when dealing with multi-domain translation.  ... 
arXiv:1811.02074v1 fatcat:al26lwojuvbkbi2dviyls4zaye

A Multi-Domain Collaborative Transfer Learning Method with Multi-Scale Repeated Attention Mechanism for Underwater Side-Scan Sonar Image Classification

Zhen Cheng, Guanying Huo, Haisen Li
2022 Remote Sensing  
Using different characteristics of multi-domain data to efficiently capture useful features for the sonar image classification, MDCTL offers a new way for transfer learning.  ...  In this paper, a multi-domain collaborative transfer learning (MDCTL) method with multi-scale repeated attention mechanism (MSRAM) is proposed for improving the accuracy of underwater sonar image classification  ...  • The randomly generated samples with consistent distribution of the training dataset are created by the generative adversarial networks (GAN), which are trained to learn an image-translation from low-complexity  ... 
doi:10.3390/rs14020355 fatcat:wu734pxal5hrfd4fh4txeow7cq

Deep-sea Nodule Mineral Image Segmentation Algorithm Based on Pix2PixHD

Wei Song, Haolin Wang, Xinping Zhang, Jianxin Xia, Tongmu Liu, Yuxi Shi
2022 Computers Materials & Continua  
The model uses a coarse-to-fine generator composed of a global generation network and two local enhancement networks, and multiple multi-scale discriminators with same network structures but different  ...  It is important for expanding the application of deep learning techniques in the field of deep-sea exploration and mining.  ...  Acknowledgement: Thanks to other teachers and students in the Media Computing Laboratory of the Minzu University of China and anonymous reviewers for their valuable comments and contributions to this research  ... 
doi:10.32604/cmc.2022.027213 fatcat:tjnkzgrzmnabxjiku6vflwfsz4

SDTGAN: Generation Adversarial Network for Spectral Domain Translation of Remote Sensing Images of the Earth Background Based on Shared Latent Domain

Biao Wang, Lingxuan Zhu, Xing Guo, Xiaobing Wang, Jiaji Wu
2022 Remote Sensing  
The introduction of shared latent domain allows multi-spectral domains connect to each other without the need to build a one-to-one model.  ...  Based on the shared latent domain hypothesis and generation adversarial network, this paper proposes the SDTGAN method to mine the correlation between the spectrum and directly generate target spectral  ...  Informed Consent Statement: Not applicable. Data Availability Statement: No new data were created or analyzed in this study. Data sharing is not applicable to this article.  ... 
doi:10.3390/rs14061359 fatcat:jbgcexogl5efdbdjwo4ggmulvy

Retrieval Guided Unsupervised Multi-domain Image-to-Image Translation [article]

Raul Gomez, Yahui Liu, Marco De Nadai, Dimosthenis Karatzas, Bruno Lepri, Nicu Sebe
2020 arXiv   pre-print
However, synthesizing new images is extremely challenging especially in multi-domain translations, as the network has to compose content and style to generate reliable and diverse images in multiple domains  ...  Image to image translation aims to learn a mapping that transforms an image from one visual domain to another.  ...  quality images from multi-modal and multi-domain image-to-image translations; • To our knowledge, we are the first to train a retrieval system exploiting image-to-image translation model generated images  ... 
arXiv:2008.04991v1 fatcat:q3kkauvhhzd23pqvkprt3ycwvi

DeepEDN: A Deep Learning-based Image Encryption and Decryption Network for Internet of Medical Things [article]

Yi Ding, Guozheng Wu, Dajiang Chen, Ning Zhang, Linpeng Gong, Mingsheng Cao, Zhiguang Qin
2020 arXiv   pre-print
Internet of Medical Things (IoMT) can connect many medical imaging equipments to the medical information network to facilitate the process of diagnosing and treating for doctors.  ...  In order to facilitate the data mining directly from the privacy-protected environment, a region of interest(ROI)-mining-network is proposed to extract the interested object from the encrypted image.  ...  The encryption network G attempts to generate an images G(x) similar to the image in domain Y , while the discriminator network D aims to find the difference between translated samples from G(x) and real  ... 
arXiv:2004.05523v2 fatcat:bck62nothzftzeqfkl7g6xq2ye
« Previous Showing results 1 — 15 out of 30,521 results