File uploads: We have fixed an issue which caused file uploads to fail. We apologise for the inconvenience it may have caused.

Published November 2, 2023 | Version v1
Dataset Open

COCO-10

  • 1. ROR icon Fondazione Bruno Kessler
  • 2. ROR icon Free University of Bozen-Bolzano

Description

COCO-10 is a database of 5700 color images (with masks) displaying 150 printed color barcodes acquired by different smartphone cameras under different illuminations. The total number of images in COCO-10 is 11700.

COCO-10: COlour BarCOde data-set contains 5700 colour images depicting 150 colour barcodes printed on two different kinds of paper and acquired under different illuminations by different smartphone cameras. Masks specifying the position of the barcodes in the acquired images are also provided. The number 10 in the data-set name refers to the fact that the colours of the barcode lines have been randomly picked up from a palette of 10 colours, including both warm and cold hues. The images of the 150 colour barcodes created from black & white barcodes are released too, along with a set of 150 synthetic colour images, where colour barcodes have been super-imposed over cluttered, real-world backgrounds. The total number of images in COCO-10, including masks, is 11700.

Aim: COCO-10 has been specifically designed for developing and/or testing algorithms for device- and illuminant- invariant colour barcode detection and decoding. It can be used also for developing and testing algorithms for gamut and tone mapping, colour correction, machine colour constancy. 

Structure: COCO-10 is organized in three folders: COLOUR-BARCODES (150 images), COLOUR-BARCODES-ON-WHITE-PAPERS (10800 images) and COLOUR-BARCODES-ON-CLUTTERED-BACKGROUND (750 images).

Short Description: Colour barcodes have been generated by coloring the lines of 150 black & white barcodes (see folder COLOUR-BARCODES). These barcode images have been printed on white papers with different density (i.e., 80 gr/m2 and 160 gr/m2) and captured by three smartphone cameras under six illuminants (i.e., a natural light and five artificial lights), yielding 5400 images of colour barcodes on uniform background (see folder COLOUR-BARCODES-ON-HITE-PAPERS, subfolders named CAMERA-X-PAPER-Y, where X and Y denote respectively the camera and the paper density). Barcodes have been also synthetically super-imposed on complex backgrounds, yielding 150 colour images (see folder COLOUR-BARCODES-ON-CLUTTERED-BACKGROUND, subfolder SYNTHETIC-COLOUR-BARCODES-ON-CLUTTERED-BACKGROUND). These latter ware divided into six subgroups,  each of which printed on sheets with density either 80 gr/m2 or 160 gr/m2 and acquired by two smartphone cameras under one of the six lights mentioned before, providing 300 images where colour barcodes appear on clutter backgrounds (see folder COLOUR-BARCODES-ON-CLUTTERED-BACKGROUND, subfolders named CAMERA-X-PAPER-Y-LIGHT-Z, where X, Y, Z indicate respectively the camera, the paper density and the light used in that acquisition) . For these acquisitions, six smartphone cameras were used, including those employed for capturing the barcodes printed on white paper. For all the 5700 acquired images we provided masks, i.e., binary images specifying the position of the barcode in the image. Therefore, the total number of images in COCO-10 is 11700.

 

Files

COCO-10.zip

Files (15.9 GB)

Name Size Download all
md5:608a568c97f3a64efd01068fa3b1930a
15.9 GB Preview Download

Additional details

Dates

Available
2023-11-06
Image dataset

References

  • M. Lecca, P.Lecca, 'A Dataset for Illuminant- and Device- invariant Colour Barcode Decoding with Cameras', Data in Briefs, To Appear