EvaluateAsync method for the Windows Machine Learning, EmotionRecognition sample.

EmotionRecognizer.part3.cs

public class EmotionRecognizer
{
    // .... 
    
    public async Task<Results> EvaluateAsync(VideoFrame inputFrame)
    {
        if (model != null) {
            // Create bindings for the input and output buffer
            var binding = new LearningModelBindingPreview(model as LearningModelPreview);
            var outputVariableList = new List<float>();
            binding.Bind(inputImageDescriptor.Name, inputFrame);
            binding.Bind(outputTensorDescriptor.Name, outputVariableList);

            // Process the frame through the model
            var results = await model.EvaluateAsync(binding, String.Empty);
            var resultProbabilities = results.Outputs[outputTensorDescriptor.Name] as List<float>;

            // return the result
            return new Results()
            {
                Neutral = resultProbabilities[0],
                Happiness = resultProbabilities[1],
                Surprise = resultProbabilities[2],
                Sadness = resultProbabilities[3],
                Anger = resultProbabilities[4],
                Disgust = resultProbabilities[5],
                Fear = resultProbabilities[6],
                Contempt = resultProbabilities[7]
            };
        }

        return null;
    }
}