CIB deepER enables the precise extraction of text from images and scanned business documents. Integration into automated and process-controlled dark processing increases the efficiency and accuracy of work processes.
Whether curved or twisted text, book page or invoice, print or handwriting - CIB deepER has been trained for excellent OCR quality.
CIB deepER delivers highly reliable results, especially for invoices and other typical documents; here too, integration into automated processes is optimal.
We train the AI model to your specific requirements. In this way, you have your own AI ready for use in just a few days, and it is constantly learning from then on.
Our text recognition works with multilingual or foreign language texts. CIB deepER supports German, English, Spanish and Portuguese. Further languages and fonts can be easily trained.
CIB deepER works with any image or PDF format in a wide range of sizes, aspect ratios and resolutions.
Our OCR integrates with your applications to provide automated, real-time text recognition.
Classic OCR systems are supported by dictionaries or LLMs. This method works well for recognising the meaning of a text.
Accurate recognition of numbers is essential for business documents. This is where dictionary-based OCR systems fail - because numbers defy the use of dictionaries.
CIB deepER achieves its recognition rates using an AI system without dictionaries and achieves particularly good results in this area.
The detection performance is consistently high, even with numbers or codes, and can easily be optimised by the customer through training for special cases!
CIB deepER demonstrated in comparison with ABBYY FineReader Tesseract 3, 4, and Omnipage Capture, "significantly better recognition results with comparable execution time“
Source: Effiziente, lexikonfreie OCR mit Deep Learning, Fraunhofer Institut, ICDAR
Integrate CIB AI into your software and offer your customers even more efficient solutions.
Specialized in business documents in German and European languages
Our AI technology has an impressive database with 33.6 million lines of training documents. This enables us to process very large models and training data sets internally.
Customised AI in just three days:
After a few days of intensive training we provide you with your own personalised AI model.
NVIDIA Hopper GPU Architecture and H100
The DeepER research project was carried out in cooperation with Fraunhofer IAIS and funded by the KMU-Innovativ programme of the German Federal Ministry of Education and Research.
Erfahren Sie mehr über unsere hocheffiziente OCR-Technologie mit künstlicher Intelligenz auf Wikipedia.
| BASIS | BUSINESS | |
|---|---|---|
|
One of the world's first AI-based text recognition |
|
|
|
Text recognition for numerous document formats |
|
|
|
Documents and texts cannot be viewed by third parties |
|
|
|
Text recognition for documents with more than 10 pages |
|
|
|
API access |
|
|
