marcosv/rawir

这是一个包含不同智能手机（如iPhoneX、SamsungS9、Samsung21、Google Pixel 7-9、Oppo vivo x90）拍摄的RAW图像的数据集。所有图像都遵循Bayer Pattern的RGGB格式，并已经过白平衡校正。每个RAW图像被分割成多个512x512x4大小的块，RGB图像则是手机成像管道（ISP）的输出，以无损PNG格式保存。场景包括室内/室外、白天/夜晚、不同的ISO级别和不同的快门速度级别。数据集可用于多种任务，如从sRGB图像重建RAW图像、学习ISP处理RAW图像、给RAW图像添加噪声并训练去噪器等。

This dataset includes RAW images captured from different smartphones such as iPhoneX, SamsungS9, Samsung21, Google Pixel 7-9, Oppo vivo x90. All images follow the RGGB format of the Bayer Pattern and have been white-black level corrected. Each RAW image is split into multiple patches of size 512x512x4, and the RGB images are the outputs of the phone imaging pipeline (ISP), saved in lossless PNG format. Scenes include indoor/outdoor, day/night, different ISO levels, and different shutter speed levels. The dataset can be used for various tasks such as reconstructing RAW images from sRGB counterparts, learning an ISP to process RAW images, adding noise to RAW images and training a denoiser, etc.