GermanyFinance and insurance2022 – ongoingDebeka Group
Combined text and image classification for incoming documents
Debeka's supplier reported manual document indexing fell below 10% in its building-savings division after adding image classification
Problem
Debeka's incoming documents included accident photographs and property plans that text classification handled poorly. Image-heavy documents in its vehicle-services and building-savings divisions had required extensive manual indexing.
Approach
Debeka and Insiders Technologies added deep-learning image classification to the existing smart FIX text classifier. The Cognitive Classifier selected text recognition, image recognition or a combination for each document. Existing workflows and the staff interface stayed in place.
Results
Takeaway
Debeka added image classification within its existing document-routing process. The reported manual-indexing reduction concerned its building-savings division.
Categories
- Size
- Large enterprise
- Industry
- Finance and insurance
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer visionPredictive analytics and ML
- Adoption
- Partner co-development
- Evidence
- Vendor case study
Sources
Source summaries
Insiders Technologies described the deployed combined classifier. It reported that manual indexing fell below 10% in Debeka's building-savings division and explained that existing workflows remained unchanged.
Full details are in the original sources.
