A deep neural network engine designed specifically for embedded applications. Powerful yet compact. Input and output are additionally processed to improve accuracy, reduce footprint and increase speed.

Learn More

byET-engine   fast and small deep neural networks for embedded systems

byET-engine is a deep neural network engine designed for embedded systems. It is lightweight and fast, making efficient use of the limited resources available. The engine itself uses only a few kilobytes of memory and employs different techniques to reduce the footprint of the deployed DNN. It requires no external libraries and can run directly on bare-metal hardware, an RTOS or a full OS. It can be used to deploy DNNs to deeply embedded systems like ARM Cortex M, Renesas RX, Microchip dsPIC and others.

byET-engine uses DNNs directly avoiding load overhead and duplication of data on the target. The main body of the software is written in portable C, with platform specific customizations in C and assembler to use the available hardware acceleration capabilities. It supports many neural network architectures: FC, CNN, LSTM, GRU.The input and output are processed using proprietary technology, to reduce the amount of resources necessary to achieve the desired results.

A DNN can be trained on a server with byET-engine, where it will use the CPU or GPU if available, or developed and trained using one of the popular deep learning frameworks and later transformed to the byET-engine format for deployment. byET-engine can be used for inference only or for continued training of a pre-trained DNN when deployed on the embedded device.

byET-engine use cases

byET-engine can be bundled with several components into one integrated system. At the heart of this software packages is our powerful yet compact neural network inference engine. Input and output are additionally processed to improve accuracy and speed.


AI software package to be used in intelligent customer specific sensor systems, giving them the power of machine learning. Use our solution to discover the function that you want to match. Deploy your trained model to your embedded device for inference. Our package is fully cutsomizeable and can be used either memoryless or time based.


AI software package with zero copy architecture for speech recognition, which allows you to add speech command recognition to your embedded device. It includes a predefined base repertoire of verbal commands and can be adapted to the customer needs in size/speed and language.


AI software package, optimized for image recognition (classification / object detection). It provides flexible input e.g. camera, touch, sensor and supports any image format. Also here product specific adaptations are possible.

we can help. get in touch with us