A software package for automatic checking of handwritten works

Subject of research: the time spent by teachers at educational institutions during the verification of handwritten assignments (tests, independent assignments), as well as algorithms for automatic recognition and verification of handwritten text.

Purpose of research: development a software for verifying handwritten assignments of students in general education organizations using optical character recognition (OCR) algorithms, machine learning, and web development tools.

Objects of research: processes and methods for automated verification of handwritten assignments of students in general education organizations using modern text recognition and machine learning technologies.

Research methods: analysis of handwriting recognition methods and techniques during the Unified State Exam, a review of modern transformer-type OCR services and frameworks, and the construction of a client-server web application architecture.

Research findings: the developed architecture of the software package, including modules for model initialization, image preprocessing with alignment, noise reduction, and segmentation functions, text recognition, results aggregation, and a user-friendly web interface for uploading assignments and viewing results.