README.md

metadata

license: apache-2.0

Document OCR Training Dataset

Dataset Description

This dataset contains training data for the document-ocr-insurance model, designed for optical character recognition (OCR) on insurance documents.

Dataset Summary

The Document OCR Training Dataset is a collection of insurance document images paired with their text transcriptions. It is used to train and evaluate OCR models specifically tailored for insurance industry documents.

Supported Tasks

• Image-to-Text: Extract text from insurance document images
• Document Understanding: Parse and understand insurance forms, claims, and policies

Languages

• English (en)

Dataset Structure

Data Fields

• image: The insurance document image
• text: The ground truth text transcription
• document_type: Type of insurance document (claim, policy, form, etc.)

Usage

from datasets import load_dataset

dataset = load_dataset("gcc-insurance-ml-models/document-ocr-training")

License

Apache 2.0