I-HAZE is a dataset about blurry indoor images. This dataset explores how to effectively remove blur from indoor images. Unlike most existing deblurring datasets, the blurry images in this dataset are generated using a professional fogging machine. To simplify color calibration and improve the evaluation of the deblurring algorithms, each scene contains a MacBeth color checker. (Both non-blurred and blurred images are captured under the same lighting conditions) The dataset includes 35 pairs of blurry and corresponding non-blurred indoor images.