Evaluating an ONNX model with WinML

In this section, we're going to start with an existing ONNX model. On the internet, in fact, you can find many pretrained models for some common scenarios, such as object detection or face analysis. A good and reliable source for these models is supported directly by the ONNX community, which curates an official repository on GitHub at https://github.com/onnx/models.

The model we're going to use in our application is for object recognition and it's called SqueezeNet. Given an image, it's able to recognize objects from 1,000 different categories. It's a great fit for edge scenarios since it provides reliable results with compact size; the model, in fact, weighs only 4 MB, leading to a very small increase in the size of the application. We can download the model directly from the URL https://github.com/onnx/models/blob/master/vision/classification/squeezenet/model/squeezenet1.1-7.onnx.

Thanks to this model, we're...