Browse Datasets

African Medical Records (OCR)

Nigerian Handwritten Clinical Records for OCR

ocr computer-vision clinical-text-extraction <1K records CC-BY 4.0
Access Dataset

Description

Paired images and digital ground-truth text mapping diverse Nigerian clinical handwriting, built to support handwriting OCR and clinical text extraction in healthcare.

Quick Start

Python 
from datasets import load_dataset

dataset = load_dataset("Nigeria-Health-data-OCR-pipeline/African-Medical-Records", split="train")
print(dataset[0])

Project Ideas

Prescription Pipeline

Build an OCR-to-LLM pipeline that reads a handwritten prescription image and extracts drug and dosage.

Advanced

Handwriting OCR

Fine-tune a vision-text model on the clinical handwriting and report character error rate.

Advanced

Ethical Considerations

  • Handwriting carries biometric nuance; ensure complete de-identification of patient names.