Talk

Quality Control with Models in the Loop

Date
Tuesday, November 10, 2026
Where
Online — Digital Object Authenticity Working Group (DOAWG)

In this talk, we will discuss how we can leverage machine learning to do quality control checks either in real-time or shortly after. We will discuss the machine learning framework we have designed for cultural heritage institutions, CHAI. We will look closely at our current CHAI workflow for identifying the type of image in a collection of scans and routing it to the appropriate frontier and open-source AI models for OCR/HTR, metadata extraction, and alt-text generation.

Presented to the Digital Object Authenticity Working Group (DOAWG), an interdisciplinary group of GLAM professionals and vendors working on descriptive and technical process metadata for digital surrogates of archival objects.