OID-RW, MDD400

OID-RW是一个基于真实世界的老照片数据集，包含82组图像，其中22组为字符类型，60组为风景和建筑。该数据集通过手动填充像素和使用不同相机拍摄两种方法构建，旨在帮助历史学家恢复历史并让普通人回忆过去。MDD400是一个多模态退化数据集，用于解决实际图像退化场景中的超分辨率重建问题。该数据集通过插值算法、CNN网络、GAN网络和不同比特率的视频捕捉四种方法模拟退化图像的生成过程。这两个数据集的应用领域是提高超分辨率模型的泛化能力和鲁棒性，解决真实世界图像退化问题。

OID-RW is a real-world vintage photo dataset containing 82 image groups, among which 22 groups are character-type images and 60 groups are landscape and architecture images. This dataset is constructed via two approaches: manual pixel filling and capturing with different cameras, aiming to assist historians in restoring history and enabling ordinary people to recall the past.

MDD400 is a multimodal degradation dataset designed for super-resolution reconstruction in real-world image degradation scenarios. This dataset simulates the generation process of degraded images through four methods: interpolation algorithms, Convolutional Neural Networks (CNNs), Generative Adversarial Networks (GANs), and video capture with different bit rates.

The application scenarios of these two datasets focus on improving the generalization ability and robustness of super-resolution models, and tackling real-world image degradation problems.

搜集汇总

背景与挑战

背景概述

该数据集包括两个部分：OID-RW是一个真实世界黑白老照片超分辨率数据集，包含82组图像，通过手动填充像素和不同相机拍摄构建，用于提升超分辨率模型的鲁棒性；MDD400是一个多模态退化数据集，通过插值算法、CNN网络、GAN网络和不同比特率视频捕获四种方法模拟图像退化，旨在解决真实场景中的超分辨率重建问题。实验证明，使用这些数据集训练的模型具有更好的泛化能力和纹理保持能力。