See the Wiki for more details.
The data directory contains quite a few large files, so be sure to use a stable connection when checking out the entire repository.
There are four new datasets containing similarity cores between images in the Dresden image database. These sets are named after one of the cameras in their respective datasets: pentax, praktica, canon, and fuji.
To check out the canon and fuji sets you may need to first install git large file storage, see git lfs. Also you will need bzip2 to unpack the files.