AI-enabled mobile app to recognize fingerprint patterns

Type Machine learning algorithm, Mobile application
AREAS OF EXPERTISE Computer vision, Machine learning, mobile UI/UX, camera API
TEAM 1 PM, 5 developers, 1 analyst, 1 UX/UI and 2 QA experts

This application was developed as a tool of self-knowledge and self-expression with the help of the theory of personal and professional classifications from dactyloscopy print.

The theory says that people with a specific personality type have a repetitive combination of patterns of dactyloscopy prints.

The developed application helps Customer – the company which is a member of European Association of Biometrics, to get biometric data (processed photos of fingerprints) necessary for analysis and further research easily and comfortably for its clients. The classical approach to gathering biometric data meant the usage of peripherals such as the fingerprints scanner, the software developed explicitly for this task.

iOS application

Android application

With this mobile application, gathering biometric data is much quicker, and it is sufficient to have a smartphone device with access to the internet. The reason is that taking fingerprints and the necessary postprocessing of the pictures is automated and designed for a broad audience. 

The readymade set of edited pictures can be sent to the company’s consultant for a more detailed analysis of peculiarities of character or used for creative self-expression.

  1. The task for automatisation of the process of gathering biometric data. The customer did not want the App's usability to be limited to some mobile devices. This constituted a significant challenge due to the differences in the devices' technical camera conditions and qualities on the two platforms: iOS and Android.
  2. Features of anthropometric and physiological characteristics of users of different age (small children and older people have dactyloscopy images that are hard to recognise; also some people of different age have unclear dactyloscopy images; scars, burns, cuts can influence the clarity of the image and accuracy of classification)
  1. Screen capture is automatically done when the centre of the camera's coordinations (centre of the camera is the limited area in the centre of the screen where the picture of the upper phalanx is supposed) match with the point in the centre of the upper phalanx. The search of upper phalanx borders and its central point is done by computer vision technology. The result of the capture is put into a repository for temporary files in the mobile device's main memory. As a result of the algorithm work, we have an original picture captured from the camera converted into a black-and-white outline and cut of empty space in the frame. This result is given to the validation module to check if there is a dactyloscopy image on it. The validation module also edits the image to optimise its recognition (extra cut of empty space, check for the light-struck frame, for the blur of the image, for the background is plain).Thanks to the implemented algorithm of processing and the validation module, we eliminate the possibility of irrelevant and nonvalid images getting into the neuronet, giving us a high level of correct recognition of groups of dactyloscopy images.
  2. As a result of the work on the application, we got a set of technical solutions, including the algorithm of processing the fingerprint image incoming from the mobile device camera for pre-validation, improving the quality of the original image and level of recognition of dactyloscopy image by trained neuronet. Considered design of the application and thoroughly constructed way of interaction makes it impossible for the user to get lost in the application's navigation despite the trivia task. It provides a fast and comfortable way of gathering biometrical data for users of all ages and all physical abilities. To get biometric data users need only to put fingers of the left-hand and the right-hand one by one in front of the camera (application has a function of skipping some fingers in case of missing or damaged dactyloscopy image). The user can stop the process at any moment and come back to it later.

The application's primary business task is the popularisation of the theory of classification of biometrical images into great public awareness and promotion of the Mymarq Personality Test (MPT) and the consulting and coaching services developed by the Customer's company.

In the future, we are planning to develop a unique ecosystem consisting of web-portal, web-service with access through an open API and separate mobile applications for different user needs and functions. That can help expand the company's services to corporate consulting and scale up the research of the individual biometrical characteristics.

AI algorithm for Tacticool game bots
AI algorithm for Tacticool game bots
PoC for Football analytics system
PoC for Football analytics system