Synthetic Barcode Datasets (SBD)

Overview

This barcode datasets contains:

• 100,000 LR synthetic barcode datasets along with their corresponding bounding boxes ground truth masks.
• 100,000 UHR synthetic barcode datasets along with along with their corresponding bounding boxes ground truth masks.

Size

The LR datasets (~2.2G) could be downloaded here at BarcodesLR. It constitutes images of resolution 400 x 400 px.

The UHR datasets (~150G) could be downloaded in 15G increments at BarcodesUHR00, BarcodesUHR01, BarcodesUHR02, BarcodesUHR03, BarcodesUHR04, BarcodesUHR05, BarcodesUHR06, BarcodesUHR07, BarcodesUHR08, BarcodesUHR09, and recombined. It constitutes images of resolution ~ >= 10k x 10k px.

How to recombine UHR datasts images

1. Download all the individual UHR files into the same directory (ex. myDirectory/).
2. In terminal, navigate the directory location and run the Linux command: "cat UHR* > UHR.tar.gz".
3. Decompress the recombined file with the Linux command "tar -xvzf UHR.tar.gz -C myDatasetsDirectory".

Types of barcodes

Code 39, Code 93, Code 128, UPC, EAN, PD417, ITF, Data Matrix, AZTEC, and QR among others.

Reference

The paper titled "Fast, Accurate Barcode Detection in Ultra High-Resolution Images" by Quenum et al. introducing this dataset was accepted in IEEE International Conference on Image Processing, September 2021, USA and coud be found here.

Sample synthesized UHR barcode image

Citation

J. Quenum, K. Wang and A. Zakhor, "Fast, Accurate Barcode Detection in Ultra High-Resolution Images," 2021 IEEE International Conference on Image Processing (ICIP), 2021, pp. 1019-1023, doi: 10.1109/ICIP42928.2021.9506134.