In image collections, the same object may appear multiple times in different photographs, crops or registrations. Fragments of objects, for example damaged or incomplete works, may also be recorded separately from the original whole. As a result, relationships within the collection can disappear from view. With Matching, we help make these visual similarities visible. By comparing images, duplicates can be identified and fragments can be linked to complete objects.

We help detect duplicate images and connect visual fragments to complete objects. We compare image material on the basis of visual characteristics, crops, composition and other similarities. This makes it visible when the same photograph occurs more than once, when different registrations relate to the same object, or when a fragment can be connected to a complete work. This is valuable for damaged objects, detail images and dispersed visual material. Substantive review always remains with your team.

Control and quality

AI does the preparatory work. Results remain reviewable and are configured around your sources, fields, terminology and quality criteria. Where relevant, the source or passage remains traceable. Your organisation decides what is adjusted, approved and ultimately used.

A concrete first step

We prefer to start with a clearly defined test set. Together we agree the required output, quality criteria and review flow. This shows what works before you scale up.