DENKblatt

‘The AI analysis with DENKweit provides us with a new depth of quality control - fast, scalable and reliable’ - REC Solar Holdings

REC Solar Holdings has fundamentally advanced its electroluminescence (EL) inspection - with image analysis AI from DENKweit, they are optimising the quality assurance of their solar cell production.

With DENKnet, EL images are analysed and even the finest defects such as microcracks, dark areas or printing defects are detected and automatically categorised into different classes - far beyond a simple good/bad assessment. 

With automated classification, heatmap visualisation of recurring defects and seamless database integration, the causes of defects can now be systematically identified and rectified in the long term.

The analysis runs 24/7 - quickly, consistently and precisely.

By integrating this technology, REC is strengthening its leading role in solar technology and setting new standards in terms of quality and process intelligence.

You can find the full story here: REC customer story 

How can you get started quickly with AI-supported image analysis?

In our live event, we will show you how to use ‘1-Class Classification’ to create an initial AI model in just a few steps, distinguish between ‘good’ and ‘bad’ and optimise it on an ongoing basis - without spending a lot of time and data.

Ideal for anyone who wants to integrate Vision AI into their processes simply but effectively. 🚀

📅 Date: 2 June 2025
🕒 Time: 3.30 pm
🎙️ With Arvid Moritz

➡️ Where? On LinkedIn! Register quickly for the free event!

Info: Event is in German. Write to us if you are interested in the English presentation.

You can now find our proven text recognition technology DENK-OCR in the ‘Pre-Trained Models’ section of the DENK Vision AI Hub.

The reason: With our DENK-OCR, we provide you with a pre-trained neural network that is immediately ready for use in many applications - especially for easily legible texts such as cleanly printed fonts. No additional training is required for most standard cases.

If you are working with more difficult conditions - such as heavily soiled text, individual handwriting or extreme lighting conditions - you can continue to retrain the model with your own images in order to achieve optimum recognition rates.

In the ‘Pre-Trained Models’ section, you will also find our ‘Licence Plate Detection’ and ‘Face Detection’, which you can use to get started straight away in many cases - without any additional training effort and, as always, without any AI knowledge.

New location, proven performance - DENK-OCR is now available under ‘Pre-Trained Models’ in the Vision AI Hub.

As always, we are happy to answer any questions you may have: support@denkweit.de.