Upstage Supplies OCR Pack to Hanwha Life
Key point
Upstage is supplying its OCR Pack to Hanwha Life to strengthen automation of insurance claim documents.
Details
Upstage is supplying its optical character recognition (OCR) solution OCR Pack to Hanwha Life, a leading domestic life insurer, for the first time. The no-code/low-code AI solution will be applied to efficiently process 5 types of insurance claim documents, including medical expense receipts.
AI adoption is rapidly spreading across the financial sector, and Korea's financial AI market is projected to reach KRW 3.2 trillion by 2026. Among these technologies, OCR is a core technology that reads required items from electronic documents and various forms, and is gaining attention as the foundation for automating insurance and banking operations.
Upstage emphasized its performance based on achievements including papers at international AI conferences and Kaggle gold medal results. OCR Pack highlights the following advantages:
- A no-code/low-code UI that processes various documents with a single click
- Cost and time savings of up to 82% compared to existing manual work
- Reduces the cost for financial companies to directly develop and operate OCR models to one-tenth
- Provides pre-training models, enabling performance to be secured with as little as one-quarter of the data required by global competitors' general-purpose models
During Hanwha Life's verification process, the solution showed superior performance compared to competitors across various documents including medical expense receipts and itemized billing statements, and achieved a recognition rate of over 95% with the base model alone, according to the company. Hanwha Life plans to accelerate digital innovation by expanding its use of unstructured image data, and Upstage is broadening its scope of application beyond OCR Pack to include offerings such as its Recommendation AI Pack.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.