Image Search with Embedding Models using ANTELOPE
| MediaWiki Users and Developers Conference Fall 2025 | |
|---|---|
| Image Search with Embedding Models using ANTELOPE | |
| Talk details | |
| Description: | Showcase: Search and categorize Images in Mediawiki using the ANTELOPE docker service |
| Speaker(s): | Kolja Bailly |
| Type: | Talk |
| Audience: | Developers, Community, Business people, Academics, Admins |
| Length: | 15 minutes |
| Video: | not available |
| Keywords: | AI, ANTELOPE, Images, Wikibase4Research |
| Give feedback | |
The Open Science Lab(OSL) at TIB Hannover, develops open source solutions for the management of research data with the MediaWiki software suite. This presentation shows a first approach of integrating the data annotation service “ANTELOPE” into mediawiki. Featuring Antelope’s image classification functions based on vector embedding models like OpenAIs CLIP model.
Using Antelope, images in Mediawiki can be searched using natural language queries or predefined categories. The Service also offers functions to add mediawiki categories to the image pages. This can be used to automatically categorize a large set of images, with the option to curate the categories later using existing mediawiki category tools and workflows. The new Antelope service is supported by wikibase4research, making the setup easy-to-use.
Git Repository: https://gitlab.com/nfdi4culture/wikibase4research/auxiliary-service-repositories/wikibase-antelope
Antelope Showcase: https://service.tib.eu/annotation
