CoBRA-CYL-AZ: Dataset of Aztec codes on Cylindrical Surfaces

Dataset summary for CoBRA-CYL-AZ The CoBRA-CYL-AZ dataset includes 7,176 images in total of bent and cropped Aztec codes and their precise and automatically computed annotations. There are two folders in the root directory of the dataset: images and markup. Each of them contains subfolders with files divided into categories based on barcode's type (aztec_code, aztec_compact, or cropped_aztec_compact) and rotation angle on a cylindrical surface (0, 15, or 30 degrees). Each image of the dataset contains a single Aztec code only. Distribution of images among subsets of the CoBRA-CYL-AZ dataset rotation_00 rotation_15 rotation_30 aztec_code 914 902 872 aztec_compact 754 752 738 cropped_aztec_compact 754 752 738 File naming pattern Each image is named using the following pattern: aa_bb_cc_dd_eeeee.png, where aa is a shorthand for aztec type (f for full or aztec_code, c for aztec_compact, cc for cropped_aztec_compact), bb is the barcode rotation angle, cc is an index of used 3D model and texture pair in the CoBRA barcode synthesis system, dd is an index of camera location on the predefined grid of view points, and eeeee is a unique hash of the image in the entire dataset. The corresponding ground truth for each image is stored under the same file name with the appended .json file extension. Annotation format Each JSON ground truth file provides the following information about a single Aztec in the image: barcode symbology (key symbology), always equal to AZTEC barcode dimensions in the number of modules (keys dim1 and dim2) barcode message in Base64 encoding (key value) barcode message in UTF-8 encoding (key value_utf8) barcode quadrangle (key quad) barcode grid of modules (key grid) The vertices of the quadrangle (quad) are indexed starting from the upper left corner of the code spiraling clockwise.The grid is stored as a list of barcode module corners which is ordered by rows starting from the upper left corner and lasting with its lower right one. Acknowledgements I am greatly thankful to all my friends who helped me to collect the photos of Aztec Compact codes for the Photo-AZ-CC dataset. Thank you for your enthusiasm and dedication. It is you who made this dataset real.