This dataset contains 1050 blurred and sharp images (350 triplets), each image triplet is a set of three photos of the same scene: sharp, defocused-blurred and motion-blurred images.

The dataset was created to validate the blur detection algorithm. The dataset can also be used for testing image deblurring, hovewer, the triplets are not "pixel-to-pixel" images, so, one cannot compare blurred and sharp images on the basis of PSNR or SSIM but sharp images can be used for visual comparison.

The dataset contains three folders: sharp, defocused-blurred and motion-blurred images.

The filename structure is as follows: id_device_type.extension where

• ID - a number from 0 to 349;
• device - the image capture device;
• type - one of [S, F, M]. S stands for Sharp image, F - deFocused-blurred image and M - Motion-blurred image.

The dataset contains 66 devices, these are typically smartphones, but several cameras are also provided.

Kaggle dataset (images were scaled to 2048 pixels by the widest side): https://www.kaggle.com/kwentar/blur-dataset

Google drive (source images): https://drive.google.com/open?id=1RObmCDPeQ1Lg-V6u7dT02Pf0qH-QMcTp

Mirror: https://yadi.sk/d/_Wli3wSSScnzCg

Sharp	Defocused-blurred	Motion-blurred

CC0: Public Domain

will be soon