Windows.AI.MachineLearning Namespace

Reference

Important

Some information relates to prerelease product that may be substantially modified before it’s released. Microsoft makes no warranties, express or implied, with respect to the information provided here.

Edit

Enables apps to load machine learning models, bind features, and evaluate the results.

Classes

ImageFeatureDescriptor	Describes the properties of the image the model is expecting.
ImageFeatureValue	Describes the properties of the image used to pass into a model.
LearningModel	Represents a trained machine learning model.
LearningModelBinding	Used to bind values to named input and output features.
LearningModelDevice	The device used to evaluate the machine learning model.
LearningModelEvaluationResult	Get the results of the evaluation.
LearningModelSession	Used to evaluate machine learning models.
LearningModelSessionOptions	Describes inference options that are used during the creation of LearningModelSession objects.
MapFeatureDescriptor	A map is a collection of (key, value) pairs.
SequenceFeatureDescriptor	A sequence is an array of elements.
TensorBoolean	A boolean tensor object.
TensorDouble	A 64-bit float tensor object.
TensorFeatureDescriptor	Tensors are multi-dimensional arrays of values.
TensorFloat	A 32-bit float tensor object.
TensorFloat16Bit	A 16-bit float tensor object.
TensorInt16Bit	A 16-bit signed integer tensor object.
TensorInt32Bit	A 32-bit signed integer tensor object.
TensorInt64Bit	A 64-bit signed integer tensor object.
TensorInt8Bit	An 8-bit signed integer tensor object.
TensorString	A string tensor object.
TensorUInt16Bit	A 16-bit unsigned integer tensor object.
TensorUInt32Bit	A 32-bit unsigned integer tensor object.
TensorUInt64Bit	A 64-bit unsigned integer tensor object.
TensorUInt8Bit	A 8-bit unsigned integer tensor object.

Interfaces

ILearningModelFeatureDescriptor	Describes the common properties that all features have.
ILearningModelFeatureValue	The instantiated value for a feature.
ILearningModelOperatorProvider	Describes the operaators for a learning model.
ITensor	Tensors are multi-dimensional values.

Enums

LearningModelDeviceKind	Defines the list of device kinds that can evaluate a machine learning model.
LearningModelFeatureKind	Input and output feature kinds for a machine learning model.
LearningModelPixelRange	Defines the list of image nominal pixel range suppored by Windows ML. The proper value is specified in a machine learning model's metadata.
TensorKind	Defines the list of supported tensor data types.

Examples

The following example loads a model, creates an evaluation session, gets the input and output features of the model, binds those features, and evaluates.

private async Task LoadAndEvaluateModelAsync(VideoFrame _inputFrame, string _modelFileName)
{
    LearningModel _model;
    ImageFeatureDescriptor _inputImageDescription;
    TensorFeatureDescriptor _outputImageDescription;
    LearningModelBinding _binding = null;
    VideoFrame _outputFrame = null;
    LearningModelSession _session;

    try
    {
        // Load and create the model
        var modelFile = 
            await StorageFile.GetFileFromApplicationUriAsync(new Uri($"ms-appx:///Assets/{_modelFileName}"));
        _model = await LearningModel.LoadFromStorageFileAsync(modelFile);

        // Create the evaluation session with the model
        _session = new LearningModelSession(_model);

        //Get input and output features of the model
        List<ILearningModelFeatureDescriptor> inputFeatures = _model.InputFeatures.ToList();
        List<ILearningModelFeatureDescriptor> outputFeatures = _model.OutputFeatures.ToList();

        // Retrieve the first input feature which is an image
        _inputImageDescription =
                inputFeatures.FirstOrDefault(feature => feature.Kind == LearningModelFeatureKind.Image)
                as ImageFeatureDescriptor;

        // Retrieve the first output feature which is a tensor
        _outputImageDescription =
                        outputFeatures.FirstOrDefault(feature => feature.Kind == LearningModelFeatureKind.Tensor)
                        as TensorFeatureDescriptor;

        //Create output frame based on expected image width and height
        _outputFrame = new VideoFrame(
            BitmapPixelFormat.Bgra8, 
            (int)_inputImageDescription.Width, 
            (int)_inputImageDescription.Height);

        //Create binding and then bind input/output features
        _binding = new LearningModelBinding(_session);

        _binding.Bind(_inputImageDescription.Name, _inputFrame);
        _binding.Bind(_outputImageDescription.Name, _outputFrame);

        //Evaluate and get the results
        var results = await _session.EvaluateAsync(_binding, "test");
    }
    catch (Exception ex)
    {
        StatusBlock.Text = $"error: {ex.Message}";
        _model = null;
    }
}

Remarks

Windows Server

To use this API on Windows Server, you must use Windows Server 2019 with Desktop Experience.

Thread safety

This API is thread-safe.