Fingerprint cards conversion into digital format

Machine learning based SDK for automated processing of paper fingerprint cards and its conversion to digital form with minimal human adjudication or correction.
Digitization of paper fingerprint cards is an actual task both for daily work of law enforcements, because they still do fingerprinting on paper, and for conversion of existing archives to digital format.
In many criminal AFIS this task is still implemented in semi-automated mode: an operator needs to enter some information manually or do manual adjustments after automated detection. In that way conversion of fingerprint cards becomes a time-consuming and tedious work, which wastes a lot of time of law enforcement officers.
Neurodactyl SDK for fingerprint cards digitization makes it easier and faster in order of magnitude. Machine learning algorithms, implemented in the SDK, analyze images similarly to human brains, accurately finding all "objects of interests" and allowing to process majority of fingerprints cards in automatic mode with minimal human adjudication.

of fingerprint cards are processing
without any manual correction

  • Detection and segmentation of flat and rolled fingerprints on an image
  • Detection of lower and upper landmarks on a fingerprint, fingerprint core, fingerprint tilt from vertical axis; provision of fully segmented fingerprint image
  • Check of position correctness for all fingerprints on a fingerprint card (rolled fingerprints and slaps)
  • Detection of missing fingers with its position
  • Detection of right and left hand
  • Validation that flat and rolled fingerprints are the same (belong to one person)
  • Any images are supported
    SDK works with any images from 150 dpi and higher. No scan modes for different types of cards are required.
  • All forms are supported
    You can work with any forms of fingerprint cards or with fingerprints on blank paper
  • Artifact-resilient
    Paper color, markup, written or printed text on a fingerprint don't affect detection and segmentation accuracy
  • Fully automated
    User doesn't need to set up input parameters or mark something manually
Examples of automatic segmentation
Do you want to find out more?
Text us here or via