Filters








17 Hits in 2.4 sec

MineGAN: effective knowledge transfer from GANs to target domains with few images [article]

Yaxing Wang, Abel Gonzalez-Garcia, David Berga, Luis Herranz, Fahad Shahbaz Khan, Joost van de Weijer
<span title="2020-04-02">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
with few target images, outperforming existing methods.  ...  We propose a novel knowledge transfer method for generative models based on mining the knowledge that is most beneficial to a specific target domain, either from a single or multiple pretrained GANs.  ...  We acknowledge the support from Huawei Kirin Solution, the Spanish projects TIN2016-79717-R and RTI2018-102285-A-I0, the CERCA Program of the Generalitat de Catalunya, and the EU Marie Sklodowska-Curie  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1912.05270v3">arXiv:1912.05270v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/go4sre7wrzgelcddjpahmkgtfe">fatcat:go4sre7wrzgelcddjpahmkgtfe</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200404000742/https://arxiv.org/pdf/1912.05270v3.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1912.05270v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

MineGAN++: Mining Generative Models for Efficient Knowledge Transfer to Limited Data Domains [article]

Yaxing Wang, Abel Gonzalez-Garcia, Chenshen Wu, Luis Herranz, Fahad Shahbaz Khan, Shangling Jui, Joost van de Weijer
<span title="2021-04-28">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
transfers knowledge to domains with few target images, outperforming existing methods.  ...  In addition, MineGAN can successfully transfer knowledge from multiple pretrained GANs.  ...  ACKNOWLEDGMENTS We acknowledge the support from Huawei Kirin Solution. We also acknowledge the help and discussion of David Berga.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2104.13742v1">arXiv:2104.13742v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/l2lyybjovrgf3bguecv2w7xhgy">fatcat:l2lyybjovrgf3bguecv2w7xhgy</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210430030855/https://arxiv.org/pdf/2104.13742v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/e1/71/e171d77caeacdb0d7a7adc7a213f4dab2da97bc6.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2104.13742v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

D3T-GAN: Data-Dependent Domain Transfer GANs for Few-shot Image Generation [article]

Xintian Wu, Huanyu Wang, Yiming Wu, Xi Li
<span title="2022-05-12">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
A typical solution for few-shot generation is to transfer a well-trained GAN model from a data-rich source domain to the data-deficient target domain.  ...  In this paper, we propose a novel self-supervised transfer scheme termed D3T-GAN, addressing the cross-domain GANs transfer in few-shot image generation.  ...  First, we transform the target data from target domain to source domain through GAN-Inversion. Then, we transfer the knowledge of the transformed data to the target domain.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2205.06032v1">arXiv:2205.06032v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/riojohj7bjcsbkvjhhurwunj4u">fatcat:riojohj7bjcsbkvjhhurwunj4u</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220514205238/https://arxiv.org/pdf/2205.06032v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/fc/c1/fcc1aaf596b1aba945c5839dd1a6e5e2e7bf8ab7.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2205.06032v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Few-shot Image Generation with Elastic Weight Consolidation [article]

Yijun Li, Richard Zhang, Jingwan Lu, Eli Shechtman
<span title="2020-12-04">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Few-shot image generation seeks to generate more data of a given domain, with only few available training examples.  ...  We demonstrate the effectiveness of our algorithm by generating high-quality results of different target domains, including those with extremely few examples (e.g., <10).  ...  We then adapt it to the target domain (e.g., Moïse Kisling faces [44] ) with just a few examples to generate more data in target domain (all images are of size 256×256).  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2012.02780v1">arXiv:2012.02780v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/b3q34zdrq5e2fevw7nhqyh2i4u">fatcat:b3q34zdrq5e2fevw7nhqyh2i4u</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210902211029/https://arxiv.org/pdf/2012.02780v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/55/84/55845cc1b81bc2c4bf8a4ba2ee8db9cbba574c8c.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2012.02780v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

A Closer Look at Few-shot Image Generation [article]

Yunqing Zhao, Henghui Ding, Houjing Huang, Ngai-Man Cheung
<span title="2022-05-08">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
However, when transferring the pretrained GANs on small target data (e.g., 10-shot), the generator tends to replicate the training samples.  ...  Modern GANs excel at generating high quality and diverse images.  ...  We conjecture that, due to the informative prior knowledge on the source domain, it is not difficult to obtain adequate realisticness when transferring to the small target domain. • Image diversity (Observation  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2205.03805v1">arXiv:2205.03805v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/q5cajk7gabdjzgczh7yp64shfa">fatcat:q5cajk7gabdjzgczh7yp64shfa</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220511040942/https://arxiv.org/pdf/2205.03805v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/30/04/300407516829a1e74a9d63dcfb4c056ac01b71d1.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2205.03805v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Few Shot Generative Model Adaption via Relaxed Spatial Structural Alignment [article]

Jiayu Xiao, Liang Li, Chaofei Wang, Zheng-Jun Zha, Qingming Huang
<span title="2022-03-31">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
A feasible solution is to start with a GAN well-trained on a large scale source domain and adapt it to the target domain with a few samples, termed as few shot generative model adaption.  ...  To relax the cross-domain alignment, we compress the original latent space of generative models to a subspace. Image pairs generated from the subspace are pulled closer.  ...  from the source domain can be well preserved and transferred to the target domain.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2203.04121v3">arXiv:2203.04121v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/lv6d6dejercvvax75axwnt67mq">fatcat:lv6d6dejercvvax75axwnt67mq</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220517030406/https://arxiv.org/pdf/2203.04121v3.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/55/c0/55c0838e3b8a8554bc4e21c4a2664b162507c616.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2203.04121v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Efficient Conditional GAN Transfer with Knowledge Propagation across Classes [article]

Mohamad Shahbazi, Zhiwu Huang, Danda Pani Paudel, Ajad Chhatkuli, Luc Van Gool
<span title="2021-03-31">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
In recent literature, it is shown that pre-trained GANs, on a different dataset, can be transferred to improve the image generation from a small target data.  ...  This motivates us to study the problem of efficient conditional GAN transfer with knowledge propagation across classes.  ...  After training the miner, Mine-GAN further fine-tunes both the generator and the miner as the final model. MineGAN is designed to transfer knowledge to a single-class target.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2102.06696v2">arXiv:2102.06696v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/ipf5666crvfppnbxehh4tsnyvm">fatcat:ipf5666crvfppnbxehh4tsnyvm</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210404171239/https://arxiv.org/pdf/2102.06696v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/b7/46/b74620bd74f62d752ca90c7015e72e9b8fa60015.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2102.06696v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Transferring Unconditional to Conditional GANs with Hyper-Modulation [article]

Héctor Laria, Yaxing Wang, Joost van de Weijer, Bogdan Raducanu
<span title="2022-04-23">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Therefore, in this paper, we focus on transferring from high-quality pretrained unconditional GANs to conditional GANs.  ...  To prevent the additional weights of the hypernetwork to overfit, with subsequent mode collapse on small target domains, we introduce a self-initialization procedure that does not require any real data  ...  We are the first to explore transfer learning from unconditional to conditional GANs, there exist only few works with which we can compare.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2112.02219v2">arXiv:2112.02219v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/5znmgefy4jdknnz6raauhnhika">fatcat:5znmgefy4jdknnz6raauhnhika</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20211208164831/https://arxiv.org/pdf/2112.02219v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/e7/06/e7068ac25aab4776dffc17c654588100e3791166.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2112.02219v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Data InStance Prior (DISP) in Generative Adversarial Networks [article]

Puneet Mangla, Nupur Kumari, Mayank Singh, Balaji Krishnamurthy, Vineeth N Balasubramanian
<span title="2021-09-21">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
We perform experiments on several standard vision datasets using various GAN architectures (BigGAN, SNGAN, StyleGAN2) to demonstrate that the proposed method effectively transfers knowledge to domains  ...  with few target images, outperforming existing state-of-the-art techniques in terms of image quality and diversity.  ...  C is a pre-trained network on a rich source domain from which we wish to transfer knowledge.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2012.04256v2">arXiv:2012.04256v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/hl4fepdoqvdfzbim4toyl27xd4">fatcat:hl4fepdoqvdfzbim4toyl27xd4</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210923095055/https://arxiv.org/pdf/2012.04256v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/30/1b/301be5628fc48a7c6d339f392c805f78875df6de.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2012.04256v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators [article]

Rinon Gal, Or Patashnik, Haggai Maron, Gal Chechik, Daniel Cohen-Or
<span title="2021-12-16">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Can a generative model be trained to produce images from a specific domain, guided by a text prompt only, without seeing any image? In other words: can an image generator be trained "blindly"?  ...  to collect even a single image.  ...  Assaf Hallak for discussions, and Zonzge Wu for assistance with StyleCLIP comparisons.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2108.00946v2">arXiv:2108.00946v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/lnn4ydsoenauxbpu6ijpm3ccn4">fatcat:lnn4ydsoenauxbpu6ijpm3ccn4</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20211224201607/https://arxiv.org/pdf/2108.00946v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/be/69/be697c79df8e4b280fec71751cb2d44667429f36.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2108.00946v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

CtlGAN: Few-shot Artistic Portraits Generation with Contrastive Transfer Learning [article]

Yue Wang, Ran Yi, Ying Tai, Chengjie Wang, Lizhuang Ma
<span title="2022-03-16">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Existing portrait stylization models that generate good quality results are based on Image-to-Image Translation and require abundant data from both source and target domains.  ...  To reduce overfitting to the few training examples, we introduce a novel Cross-Domain Triplet loss which explicitly encourages the target instances generated from different latent codes to be distinguishable  ...  MineGAN [46] proposes a miner network to find the knowledge that is most beneficial to a target domain from pretrained GANs.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2203.08612v1">arXiv:2203.08612v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/eqrot5yew5hd5iftuqofdqftpe">fatcat:eqrot5yew5hd5iftuqofdqftpe</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220324170519/https://arxiv.org/pdf/2203.08612v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/62/f9/62f91259a04f935c8a44977c096eff16356267bb.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2203.08612v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

CSG0: Continual Urban Scene Generation with Zero Forgetting [article]

Himalaya Jain, Tuan-Hung Vu, Patrick Pérez, Matthieu Cord
<span title="2022-05-02">2022</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
To this end, we introduce a novel framework that not only (i) enables seamless knowledge transfer in continual training but also (ii) guarantees zero forgetting with a small overhead cost.  ...  With the rapid advances in generative adversarial networks (GANs), the visual quality of synthesised scenes keeps improving, including for complex urban scenes with applications to automated driving.  ...  Starting from a GAN model pre-trained on previous domains, we want to reuse most of the learned weights and extend the model with small overhead (i) to leverage the knowledge learned from the previous  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2112.03252v2">arXiv:2112.03252v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/nnrigwcn6zhypg6gorqgruvdbi">fatcat:nnrigwcn6zhypg6gorqgruvdbi</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20220505192132/https://arxiv.org/pdf/2112.03252v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/c0/2d/c02d9fd1a832867d6e33506e7ddc6a45e4f57836.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2112.03252v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Sketch Your Own GAN [article]

Sheng-Yu Wang, David Bau, Jun-Yan Zhu
<span title="2021-09-20">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Finally, we demonstrate a few applications of the resulting GAN, including latent space interpolation and image editing.  ...  In this work, we present a method, GAN Sketching, for rewriting GANs with one or more sketches, to make GANs training easier for novice users.  ...  We are grateful for the support of the Naver Corporation, DARPA SAIL-ON HR0011-20-C-0022 (to DB), and Signify Lighting Research.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2108.02774v2">arXiv:2108.02774v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/cuugqdqhr5catg4uqmf7dv4ehq">fatcat:cuugqdqhr5catg4uqmf7dv4ehq</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210807062548/https://arxiv.org/pdf/2108.02774v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/97/61/976143101d894793ba6e9104c09a127fb8c66923.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2108.02774v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Data-Efficient GAN Training Beyond (Just) Augmentations: A Lottery Ticket Perspective [article]

Tianlong Chen, Yu Cheng, Zhe Gan, Jingjing Liu, Zhangyang Wang
<span title="2021-10-22">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Training generative adversarial networks (GANs) with limited real image data generally results in deteriorated performance and collapsed models.  ...  We find our coordinated framework to offer orthogonal gains to existing real image data augmentation methods, and we additionally present a new feature-level augmentation that can be applied together with  ...  Comprehensive experiments consistently demonstrate the effectiveness of our proposal, on diverse GAN architectures, objectives, and datasets.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2103.00397v3">arXiv:2103.00397v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/nx67rbomnnhbxeezreqidba6s4">fatcat:nx67rbomnnhbxeezreqidba6s4</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20211102184628/https://arxiv.org/pdf/2103.00397v3.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/6e/46/6e466c1bca96e2847be32f60677517aae628a664.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2103.00397v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

GAN Prior Embedded Network for Blind Face Restoration in the Wild [article]

Tao Yang
<span title="2021-05-13">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
with a set of synthesized low-quality face images.  ...  The GAN blocks are designed to ensure that the latent code and noise input to the GAN can be respectively generated from the deep and shallow features of the DNN, controlling the global face structure,  ...  Minegan: effective knowledge transfer from gans to target domains with few images.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2105.06070v1">arXiv:2105.06070v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/njeqkuh7pzab7btmwwdy5badiq">fatcat:njeqkuh7pzab7btmwwdy5badiq</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210623100410/https://arxiv.org/pdf/2105.06070v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/63/c7/63c778bc07ce7bcbfa8da7b45fd03948b6a663fe.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2105.06070v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>
&laquo; Previous Showing results 1 &mdash; 15 out of 17 results