Dataset

Description

The Image Dataset is a document in image form with a total of 1250 files consisting of 50 forms with the formats: png, jpeg, gif, bmp and tiff, with image sizes: 800x600, 1024x768, 1920x1080, 2560x1440 and 3840x2160.

Steps to reproduce

The method consisted in designing base certificate templates using graphic design software and then batch-processing them to produce large-scale variations. The protocol was based on generating purely simulated (dummy) data to replicate authentic credentials without containing any real personal information. They were generated by using automated Python scripts and image processing libraries (such as OpenCV or PIL) to scale the templates into five different spatial resolutions (ranging from 800x600 to 4K) and export them into five raster formats (JPG, PNG, BMP, TIFF, GIF). Here, one advantage of this dataset is that it provides a ready-made, diverse set of cover images for digital credential testing. The method is simple and highly reproducible, achieving a standardized environment for researchers to benchmark their algorithms against varying image compressions and dimensions.

Institutions

• Universitas Ahmad Dahlan

  Yogyakarta, Yogyakarta

Categories

Licence