汇总了自2024年以来新提出的高质量图像修复工作,包含扩散模型、transformer、mamba、sam等最前沿的技术,其中一些是ICLR、ICML、CVPR、ECCV、ACM MM 2024年的新作。
这里是第二部分,还有两部分请参阅。
这篇博客先做简单汇总,后续我还会逐一阅读和学习,并逐一写详细的博客介绍。
目录
Priors in Deep Image Restoration and Enhancement: A Survey
Distilling Semantic Priors from SAM to Efficient Image Restoration Models
IPT-V2: Efficient Image Processing Transformer using Hierarchical Attentions
Dynamic Pre-training: Towards Efficient and Scalable All-in-One Image Restoration
SPIRE: Semantic Prompt-Driven Image Restoration
Referring Flexible Image Restoration
Improving Image Restoration through Removing Degradations in Textual Representations
Exposure Bracketing is All You Need for Unifying Image Restoration and Enhancement Tasks
UGPNet: Universal Generative Prior for Image Restoration
BrushNet: A Plug-and-Play Image Inpainting Model with Decomposed Dual-Branch Diffusion
MxT: Mamba x Transformer for Image Inpainting
Priors in Deep Image Restoration and Enhancement: A Survey
https://github.com/VLIS2022/Awesome-Image-Prior
https://arxiv.org/pdf/2206.02070
Distilling Semantic Priors from SAM to Efficient Image Restoration Models
https://arxiv.org/pdf/2403.16368
IPT-V2: Efficient Image Processing Transformer using Hierarchical Attentions
https://arxiv.org/pdf/2404.00633
Dynamic Pre-training: Towards Efficient and Scalable All-in-One Image Restoration
https://github.com/akshaydudhane16/DyNet
https://arxiv.org/pdf/2404.02154
SPIRE: Semantic Prompt-Driven Image Restoration
Referring Flexible Image Restoration
https://arxiv.org/pdf/2404.10342
Improving Image Restoration through Removing Degradations in Textual Representations
https://github.com/mrluin/TextualDegRemoval
https://arxiv.org/pdf/2312.17334
Exposure Bracketing is All You Need for Unifying Image Restoration and Enhancement Tasks
https://github.com/cszhilu1998/BracketIRE
https://arxiv.org/pdf/2401.00766
UGPNet: Universal Generative Prior for Image Restoration
https://arxiv.org/pdf/2401.00370
BrushNet: A Plug-and-Play Image Inpainting Model with Decomposed Dual-Branch Diffusion
https://github.com/TencentARC/BrushNet
https://arxiv.org/pdf/2403.06976