【图像复原】论文精读:Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration

第一次来请先看这篇文章:【超分辨率(Super-Resolution)】关于【超分辨率重建】专栏的相关说明,包含专栏简介、专栏亮点、适配人群、相关说明、阅读顺序、超分理解、实现流程、研究方向、论文代码数据集汇总等)

文章目录

  • 前言
  • Abstract
  • 1. Introduction
  • 2. Related Work
  • 3. Method
    • 3.1. Model Scaling Up
    • 3.2. Scaling Up Training Data
    • 3.3. Restoration-Guided Sampling
  • 4. Experiments
    • 4.1. Model Training and Sampling Settings
    • 4.2. Comparison with Existing Methods
    • 4.3. Controlling Restoration with Textual Prompts
    • 4.4. Ablation Study
  • 5. Conclusion
  • Appendix
    • A. Discussions
      • A.1. Degradation-Robust Encoder
      • A.2. LLaVA Annotation
      • A.3. Limitations of Negative Prompt
      • A.4. Negative Samples Generation
    • B. More Visual Results


前言

论文题目:Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wild —— 放大到卓越:野外照片真实感图像恢复的打印模型缩放

论文地址:Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wild

项目地址:https://supir.xpixel.group/

论文源码:

你可能感兴趣的:(超分辨率重建(理论+实战,科研+应用),深度学习,人工智能,计算机视觉,图像修复,图像处理,论文阅读,论文笔记)