Fast Custom OCR for Handwritten Text on Checks

Automating the extraction of handwritten text from checks is crucial for fraud detection and efficient check processing. Although Optical Character Recognition (OCR) tools like Tesseract excel with typed text, they often falter with handwritten content, causing computational inefficiencies. This project’s focus is on developing a rapid and cost-effective OCR solution tailored for handwritten text on checks. By harnessing open-source OCR libraries alongside customized enhancements, we aim to build a robust system adept at precisely extracting names and other vital handwritten fields from checks. This technological advancement is poised to significantly improve fraud detection accuracy and streamline processing workflows in financial institutions. By overcoming the challenges posed by handwritten content variability, we anticipate enhancing the overall security and operational efficiency of financial transactions.

Faculty Supervisor:

Karteek Popuri

Student:

Partner:

NASDAQ Canada Inc

Discipline:

Engineering

Sector:

Artificial Intelligence

University:

Memorial University of Newfoundland

Program: