© NürnbergMesse, Frank Boxler

In the 'Software' category, the jury was won over by Embedded Trend and its 'byET-engine' - a DNN engine for embedded systems. It is fast, has a very small code size and uses the available resources extremely efficiently. A DNN (Deep Neural Network) can be trained with the 'byET-engine' on a server and later converted into the byET format for further use. The 'byET-engine' can already be used with the pre-trained parameters for pattern recognition or it can be further trained during the runtime of the target system in order to deliver even better recognition rates.