University of Twente Student Theses

Login

Country-independent MRTD layout extraction and its applications

Santiago Garcia, Eric (2022) Country-independent MRTD layout extraction and its applications.

[img] PDF
17MB
Abstract:Machine-readable travel documents are documents like passports or ID cards that can be used for identity verification. Current advances in AI make possible the creation of systems that use them in an automated way. Unfortunately, there isn't much work about general work with MRTDs. In part, because of the little data publicly available. The current solutions use manually-crafted templates that need to be constantly updated. Thus, the goal of this project is to find a system that can extract the template of an MRTD in a generalized way, by adapting work from other domains. Our system obtains the template of any MRTD document, by extracting the visual text fields from it and classifying each of those, giving them their label based on the location. Finally, these templates can be used to automatically delete some text fields from a document image, a common task performed by companies that need to protect the data of their customers. A second application is the generation of fake documents using the previous tools, that could be use to train or test other Machine Learning algorithms.
Item Type:Essay (Master)
Clients:
Innovalor, Enschede, Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science MSc (60300)
Link to this item:https://purl.utwente.nl/essays/93230
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page