Using existing models
Deep Neural Network (DNN) models are often complex, so in order to achieve their optimum performance, they should run in an HW accelerated chip such as GPU or DS. Due to the underlying complexity they are often also specific to certain hardware architecture. If you have an existing IoT Edge and are looking for example models, start by checking which neural networks and ML frameworks are supported.
AI models can often be converted to allow running them in the specific hardware. However, it’s extremely important to understand that the inferencing engine in the hardware needs to support also the neural network model that is used in the model.
- Use ONNX Converter Image to convert other major model frameworks to ONNX. Supported frameworks are currently
- CNTK, CoreML, Keras, scikit-learn, Tensorflow, PyTorch
Model Zoos are collections of AI models that can be run as such or improved to meet specific user’s needs.