Enhancing automatic context-driven image description for Retrieval-Augmented Generation (RAG) pipelines in the manufacturing industry.

• Unité de recherche: INNOSUISSE
• Numéro de projet: 132.030 INNO-ICT
• Titre du projet anglais: Enhancing automatic context-driven image description for Retrieval-Augmented Generation (RAG) pipelines in the manufacturing industry.

Mots-clé (Anglais)

ICT, Artificial/local/extended/networked intelligence

Résumé des résultats (Abstract) (Anglais)

The project focuses on enhancing RAG pipelines in manufacturing by improving multi-modal capabilities. Specifically, generating context-aware image descriptions from technical documents with textual design requirements, screenshots/UIs, diagrams, and technical drawings.