Automated compliance checking and approval of labels using AI
forTUN_E
Project description
An important part of product labels is the instructions for use, which are intended to ensure the proper use of the product. In addition to classic instructions for use, information on hazardous substances and associated risks is essential. Before the product is delivered, the label must therefore be carefully checked to ensure that it complies with all legal requirements.
Drafts of labels, some of which are multilingual, often contain formal or content errors. Until now, checks have been carried out by staff who manually check compliance with guidelines and layouts in order to meet safety requirements. The aim of the project is to reduce the review time and speed up the approval process by using automated review procedures based on innovative AI methods. This should result in greater process reliability, which will both reduce internal effort and improve the quality of the labels delivered.
Research contribution
As part of the project, fortiss GmbH will research suitable AI methods for the automatic extraction and comparison of textual and image-based information. In addition to classic methods such as OCR (optical character recognition), document parsing, and computer vision technologies, the use of novel LLM (large language models) and RAG (retrieval augmented generation) systems based on them will also be considered for comparing labels.
Project duration
17.11.2025 - 31.05.2026
Contact
Project partner
