Case Study: University of Oxford achieves faster, lower-cost coin image recognition and cataloging with Amazon Web Services

University of Oxford Introduces a Sector-Leading Image Recognition ML Prototype to Augment Digitization in Numismatics

University of Oxford’s Gardens, Libraries & Museums (GLAM) needed a faster way to digitize and catalog its numismatics collection, where volunteers previously spent 10 minutes to hours analyzing a single coin. Working with Amazon Web Services and using services such as Amazon SageMaker and Amazon EC2, GLAM set out to build an image recognition ML prototype to automate image processing and improve collection access.

Amazon Web Services helped GLAM build and deploy 11 machine learning models in about 10 weeks, using Spot Instances to cut training costs to 10% of on-demand pricing and reduce training time by 50%. The resulting system can correct, deblur, denoise, and tag coin images in minutes instead of hours, and is expected to save up to 3 years of work on a 300,000-coin collection while also reducing inference time from minutes to seconds.

University of Oxford

Anjanesh Babu

Systems Architect and Network Manager, Gardens and Museums IT