YUANxy2026/DPDD

DPDD数据集（双像素散焦去模糊）提供了成对的散焦和全焦图像，以及双像素子孔径视图，用于训练和评估模型在散焦去模糊任务上的性能。每个场景包括使用大光圈捕获的散焦图像、其左右双像素视图以及使用小光圈捕获的相应清晰图像。该数据集由Abdullah Abuolaim和Michael S. Brown在ECCV 2020的论文中发布，并在Hugging Face上创建了兼容版本。数据集分为训练集、验证集和测试集，分别包含350、74和76个样本。

The DPDD dataset (Dual-Pixel Defocus Deblurring) provides paired defocused and all-in-focus images, along with dual-pixel sub-aperture views, for training and evaluating models on defocus-deblurring tasks. Each scene includes a defocused image captured with a wide aperture, its left/right dual-pixel views, and a corresponding sharp image captured with a small aperture. This dataset was released with the paper **Defocus Deblurring Using Dual-Pixel Data (ECCV 2020)** by Abdullah Abuolaim and Michael S. Brown, and a Hugging Face compatible version was created. The dataset is divided into training, validation, and test sets, containing 350, 74, and 76 samples respectively.