MID dataset [Homepage]

Wenzheng Song, Masanori Suganuma, Xing Liu, Noriyuki Shimobayashi, Daisuke Maruta, Takayuki Okatani.

+ Matching in the Dark: A Dataset for Matching Image Pairs of Low-light Scenes

[Introduction] This repository contains details about the MID (Matching in the Dark) dataset. The MID dataset was introduced as a benchmark for local descriptor evaluation challenge in extreme low-light conditions. This dataset also can be used for low-light Raw image-enhancing evaluation. See the paper for more details.

• The dataset contains diverse scenes consisting of 54 outdoor and 54 indoor scenes.

• For each scene, we provide one pair of groups of multiple RAW-format images captured from different two positions.

• In each group, there are 48 (6 shutter speeds × 8 ISO settings) underexposure images and one correspond long-exposure image.

• We provide ground truth relative camera pose for each scene obtained with long-exposure images.

[Samples] Here are example stereo image pairs (long exposure versions) of four indoor scenes and four outdoor scenes!

If there is a need to manually get the MID dataset, download and untar the following file:

• MID [460GB].

Citation

If you find these models useful for your resesarch, please cite with this bibtex.