轮廓小波扩散:快速、高质量的图像生成模型

IF 1.8 4区 计算机科学 Q3 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE
Yaoyao Ding, Xiaoxi Zhu, Yuntao Zou
{"title":"轮廓小波扩散:快速、高质量的图像生成模型","authors":"Yaoyao Ding,&nbsp;Xiaoxi Zhu,&nbsp;Yuntao Zou","doi":"10.1111/coin.12644","DOIUrl":null,"url":null,"abstract":"<p>Diffusion models can generate high-quality images and have attracted increasing attention. However, diffusion models adopt a progressive optimization process and often have long training and inference time, which limits their application in realistic scenarios. Recently, some latent space diffusion models have partially accelerated training speed by using parameters in the feature space, but additional network structures still require a large amount of unnecessary computation. Therefore, we propose the Contour Wavelet Diffusion method to accelerate the training and inference speed. First, we introduce the contour wavelet transform to extract anisotropic low-frequency and high-frequency components from the input image, and achieve acceleration by processing these down-sampling components. Meanwhile, due to the good reconstructive properties of wavelet transforms, the quality of generated images can be maintained. Second, we propose a Batch-normalized stochastic attention module that enables the model to effectively focus on important high-frequency information, further improving the quality of image generation. Finally, we propose a balanced loss function to further improve the convergence speed of the model. Experimental results on several public datasets show that our method can significantly accelerate the training and inference speed of the diffusion model while ensuring the quality of generated images.</p>","PeriodicalId":55228,"journal":{"name":"Computational Intelligence","volume":null,"pages":null},"PeriodicalIF":1.8000,"publicationDate":"2024-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Contour wavelet diffusion: A fast and high-quality image generation model\",\"authors\":\"Yaoyao Ding,&nbsp;Xiaoxi Zhu,&nbsp;Yuntao Zou\",\"doi\":\"10.1111/coin.12644\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Diffusion models can generate high-quality images and have attracted increasing attention. However, diffusion models adopt a progressive optimization process and often have long training and inference time, which limits their application in realistic scenarios. Recently, some latent space diffusion models have partially accelerated training speed by using parameters in the feature space, but additional network structures still require a large amount of unnecessary computation. Therefore, we propose the Contour Wavelet Diffusion method to accelerate the training and inference speed. First, we introduce the contour wavelet transform to extract anisotropic low-frequency and high-frequency components from the input image, and achieve acceleration by processing these down-sampling components. Meanwhile, due to the good reconstructive properties of wavelet transforms, the quality of generated images can be maintained. Second, we propose a Batch-normalized stochastic attention module that enables the model to effectively focus on important high-frequency information, further improving the quality of image generation. Finally, we propose a balanced loss function to further improve the convergence speed of the model. Experimental results on several public datasets show that our method can significantly accelerate the training and inference speed of the diffusion model while ensuring the quality of generated images.</p>\",\"PeriodicalId\":55228,\"journal\":{\"name\":\"Computational Intelligence\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.8000,\"publicationDate\":\"2024-04-23\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Computational Intelligence\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://onlinelibrary.wiley.com/doi/10.1111/coin.12644\",\"RegionNum\":4,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computational Intelligence","FirstCategoryId":"94","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1111/coin.12644","RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
引用次数: 0

摘要

扩散模型可以生成高质量的图像,因此受到越来越多的关注。然而,扩散模型采用渐进优化过程,通常需要较长的训练和推理时间,这限制了其在现实场景中的应用。最近,一些潜空间扩散模型通过使用特征空间中的参数,部分加快了训练速度,但额外的网络结构仍需要大量不必要的计算。因此,我们提出了轮廓小波扩散方法来加快训练和推理速度。首先,我们引入轮廓小波变换,从输入图像中提取各向异性的低频和高频分量,并通过处理这些下采样分量实现加速。同时,由于小波变换具有良好的重构特性,可以保持生成图像的质量。其次,我们提出了批量归一化随机关注模块,使模型能有效地关注重要的高频信息,进一步提高图像生成的质量。最后,我们提出了一种平衡损失函数,以进一步提高模型的收敛速度。在多个公开数据集上的实验结果表明,我们的方法可以显著加快扩散模型的训练和推理速度,同时确保生成图像的质量。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Contour wavelet diffusion: A fast and high-quality image generation model

Diffusion models can generate high-quality images and have attracted increasing attention. However, diffusion models adopt a progressive optimization process and often have long training and inference time, which limits their application in realistic scenarios. Recently, some latent space diffusion models have partially accelerated training speed by using parameters in the feature space, but additional network structures still require a large amount of unnecessary computation. Therefore, we propose the Contour Wavelet Diffusion method to accelerate the training and inference speed. First, we introduce the contour wavelet transform to extract anisotropic low-frequency and high-frequency components from the input image, and achieve acceleration by processing these down-sampling components. Meanwhile, due to the good reconstructive properties of wavelet transforms, the quality of generated images can be maintained. Second, we propose a Batch-normalized stochastic attention module that enables the model to effectively focus on important high-frequency information, further improving the quality of image generation. Finally, we propose a balanced loss function to further improve the convergence speed of the model. Experimental results on several public datasets show that our method can significantly accelerate the training and inference speed of the diffusion model while ensuring the quality of generated images.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Computational Intelligence
Computational Intelligence 工程技术-计算机:人工智能
CiteScore
6.90
自引率
3.60%
发文量
65
审稿时长
>12 weeks
期刊介绍: This leading international journal promotes and stimulates research in the field of artificial intelligence (AI). Covering a wide range of issues - from the tools and languages of AI to its philosophical implications - Computational Intelligence provides a vigorous forum for the publication of both experimental and theoretical research, as well as surveys and impact studies. The journal is designed to meet the needs of a wide range of AI workers in academic and industrial research.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信