Computer vision for history: keyword detection in century old handwritten documents

Imagine being able to scan in minutes an archive of tens of thousands of documents for a given keyword!
This would be invaluable for archivists and historians.
This project is about creating a computer vision program that will admit as input the digital image of a handwritten word, then scan a large database containing scanned images of handwritten documents, and return the documents containing that word as well as the word's location.
 
This is a textbook object detection challenge. You are expected to use modern deep learning techniques.This project is most suitable for individuals with a keen interest in history and culture.
 
Students that have completed the EE417 (computer vision) course will be prioritized.