Government Document OCR Service
Large-Scale Document Digitization
Digitizes paper documents for government agencies with high accuracy. Large-scale parallel processing on AWS Batch can handle even millions of documents in a short time.
Overview
A service that uses OCR technology to digitize the large volumes of paper documents held by government agencies. Parallel processing with AWS Step Functions and AWS Batch efficiently handles enormous document volumes.
It uses a high-accuracy OCR engine that handles handwritten characters and old printed materials. We also built a workflow for verifying and correcting extracted data to ensure data quality.
Challenge & Solution
Large-Scale Processing
Millions of documents had to be processed within a limited timeframe.
Parallel processing on AWS Batch achieved more than 100,000 documents per day.
High-Accuracy Recognition
Recognition accuracy for handwritten characters and old printed materials was low.
AWS Textract combined with a proprietary post-processing engine achieved 99.5% accuracy.
Quality Control
Verifying and correcting OCR results required significant human cost.
We built AI-based automatic verification and an efficient correction workflow.
Tech Stack
- Next.js
- TypeScript
- AWS
- Step Functions
- AWS Batch
- Textract
- Lambda
iPaaS System