- 博客(40)
- 收藏
- 关注
原创 VICO: PLUG-AND-PLAY VISUAL CONDITION FOR PER-SONALIZED TEXT-TO-IMAGE GENERATION ——【代码复现】
VICO: PLUG-AND-PLAY VISUAL CONDITION FOR PER-SONALIZED TEXT-TO-IMAGE GENERATION ——【代码复现】
2024-07-13 20:01:38
583
原创 FreeU: Free Lunch in Diffusion U-Net——【代码复现】
FreeU: Free Lunch in Diffusion U-Net——【代码复现】
2024-07-06 21:13:21
711
原创 MasaCtrl: Tuning-Free Mutual Self-Attention Control for ConsistentImage Synthesis and Editing【代码复现】
MasaCtrl: Tuning-Free Mutual Self-Attention Control for ConsistentImage Synthesis and Editing【代码复现】
2024-07-05 21:12:17
1031
原创 Cross-Image Attention for Zero-Shot Appearance Transfer——【代码复现】
Cross-Image Attention for Zero-Shot Appearance Transfer——【代码复现】
2024-05-12 18:37:32
642
9
原创 BootPIG: Bootstrapping Zero-shot Personalized Image Generation Capabilities inPretrained Diffusion
BootPIG: Bootstrapping Zero-shot Personalized Image Generation Capabilities inPretrained Diffusion Models——【论文笔记】
2024-04-15 15:56:28
692
2
原创 Towards Efficient Diffusion-Based Image Editing with Instant Attention Masks——【论文笔记】
Towards Efficient Diffusion-Based Image Editing with Instant Attention Masks——【论文笔记】
2024-03-31 11:43:03
1053
1
原创 DreamMatcher: Appearance Matching Self-Attentionfor Semantically-Consistent Text-to-Image Personali
DreamMatcher: Appearance Matching Self-Attentionfor Semantically-Consistent Text-to-Image Personali——【代码复现】
2024-03-06 19:46:22
1019
3
原创 Uni-ControlNet: All-in-One Control toText-to-Image Diffusion Models——【论文笔记】
Uni-ControlNet: All-in-One Control toText-to-Image Diffusion Models——【论文笔记】
2024-03-05 11:03:59
2214
3
原创 DreamMatcher: Appearance Matching Self-Attentionfor Semantically-Consistent Text-to-Image Personali
DreamMatcher: Appearance Matching Self-Attentionfor Semantically-Consistent Text-to-Image Personali——【论文笔记】
2024-03-04 12:32:34
1018
原创 IP-Adapter: Text Compatible Image Prompt Adapter forText-to-Image Diffusion Models——【论文笔记】
IP-Adapter: Text Compatible Image Prompt Adapter forText-to-Image Diffusion Models——【论文笔记】
2024-02-27 11:25:58
1505
1
原创 SVDiff: Compact Parameter Space for Diffusion Fine-Tuning——【论文笔记】
SVDiff: Compact Parameter Space for Diffusion Fine-Tuning——【论文笔记】
2024-02-03 17:49:50
1992
1
原创 Encoder-based Domain Tuning for Fast Personalization of Text-to-Image Models——【论文笔记】
Encoder-based Domain Tuning for Fast Personalization of Text-to-Image Models——【论文笔记】
2024-01-29 15:29:30
1183
1
原创 Multi-Concept Customization of Text-to-Image Diffusion——【论文笔记】
Multi-Concept Customization of Text-to-Image Diffusion——【论文笔记】
2024-01-18 16:10:59
1706
3
原创 Multi-Concept Customization of Text-to-Image Diffusion——【代码复现】
Multi-Concept Customization of Text-to-Image Diffusion——【代码复现】
2024-01-13 17:32:48
1884
13
原创 Adding Conditional Control to Text-to-Image Diffusion Models——【代码复现】
Adding Conditional Control to Text-to-Image Diffusion Models——【代码复现】
2024-01-08 14:02:06
2042
18
原创 Adding Conditional Control to Text-to-Image Diffusion Models——【论文笔记】
Adding Conditional Control to Text-to-Image Diffusion Models——【论文笔记】
2024-01-06 14:54:40
1536
1
原创 Multi Diffusion: Fusing Diffusion Paths for Controlled Image Generation——【论文笔记】
Multi Diffusion: Fusing Diffusion Paths for Controlled Image Generation——【论文笔记】
2024-01-01 16:14:27
1900
1
原创 Magic Fusion: Boosting Text-to-Image Generation Performance by Fusing Diffusion Models——【论文笔记】
Magic Fusion: Boosting Text-to-Image Generation Performance by Fusing Diffusion Models——【论文笔记】
2023-12-29 12:47:15
1007
1
原创 Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models——【论文笔记】
Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models——【论文笔记】
2023-12-25 16:23:36
1667
1
原创 Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models ——【代码复现】
Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models ——【代码复现】
2023-12-24 20:16:12
1788
9
原创 Localizing Object-level Shape Variations with Text-to-Image Diffusion Models——论文笔记
Localizing Object-level Shape Variations with Text-to-Image Diffusion Models 发表于ICCV 2023
2023-12-18 17:24:40
1302
1
原创 合并排序——代码实现
合并排序(Merge Sort)是一种基于分治法(Divide and Conquer)的排序算法。其主要思想是将待排序的数组分成两部分,分别对两部分进行递归排序,然后将两个有序的子数组合并成一个有序的数组。
2023-12-13 16:35:56
124
1
原创 DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation——论文笔记
在这篇论文中,模型会出现语言漂移的原因是因为在使用少量图像进行微调时,模型可能会过度拟合这些图像,从而忘记了如何生成与这些图像相似但不完全相同的图像。这种现象被称为"过拟合",它会导致模型在生成图像时偏离先验信息,从而产生语言漂移。为了解决这个问题,DreamBooth提出了一个自生类别特定的先验保留损失,以鼓励模型生成符合先验信息的图像,从而保留先验信息的多样性和一致性。这样可以避免模型出现语言漂移现象,从而提高生成图像的质量和准确性。
2023-11-27 16:00:13
1022
2
原创 High-Resolution Image Synthesis with Latent Diffusion Models——论文解读
利用潜在扩散模型生成高分辨率的图像,文章是发表于CVPR 2022上的一篇论文
2023-11-26 21:45:24
620
1
做生成需要多大算力?
2023-11-25
TA创建的收藏夹 TA关注的收藏夹
TA关注的人