Academic
FormReader — Extracting Data from Handwritten Production Forms
A research project on extracting structured data from handwritten Turkish production forms, carried out at Sakarya University.
Research · OCR and computer vision · ongoing
PythonOpenCVPaddleOCRSVTRVision-Language Models
Problem
On production lines, downtime reasons and measurements are written onto forms by hand. Because these records never reach a digital system, they cannot be analyzed.
Solution
As a research engineer at Sakarya University, I am working on a system that extracts structured data from handwritten Turkish production forms. My M.Sc. research is on the same topic.
Technical Details
- OCR and computer vision pipelines with Python, OpenCV and PaddleOCR/SVTR-based recognition models
- Experiments with vision-language models for handwritten free-text recognition and document understanding
- Template-based document processing and geometric normalization
- Human-in-the-loop validation workflows
- Data processing, evaluation and validation pipelines for writer-independent OCR experiments
The research is supported by Sakarya University BAP (Project No. 2026-29-86-214).