public static async Task <PetModelModel> CreateFromStreamAsync(IRandomAccessStreamReference stream)
{
PetModelModel learningModel = new PetModelModel();
learningModel.model = await LearningModel.LoadFromStreamAsync(stream);
learningModel.session = new LearningModelSession(learningModel.model);
learningModel.binding = new LearningModelBinding(learningModel.session);
return(learningModel);
}
public async void Classification(byte[] bytes)
{
var file = await StorageFile.GetFileFromApplicationUriAsync(new Uri($"ms-appx:///Assets/{modelFileName}"));
petModel = await PetModelModel.CreateFromStreamAsync(file);
try
{
var newBitmap = new WriteableBitmap(255, 255);
using (var stream = new InMemoryRandomAccessStream())
{
await stream.WriteAsync(bytes.AsBuffer());
stream.Seek(0);
await newBitmap.SetSourceAsync(stream);
}
var outputBitmap = SoftwareBitmap.CreateCopyFromBuffer(
newBitmap.PixelBuffer,
BitmapPixelFormat.Bgra8,
newBitmap.PixelWidth,
newBitmap.PixelHeight
);
var frame = VideoFrame.CreateWithSoftwareBitmap(outputBitmap);
if (frame != null)
{
try
{
var inputData = new PetModelInput();
inputData.data = ImageFeatureValue.CreateFromVideoFrame(frame);
var results = await petModel.EvaluateAsync(inputData);
var loss = results.loss.ToList();
var catValue = loss.FirstOrDefault()["cat"];
var dogValue = loss.FirstOrDefault()["dog"];
var result = new Dictionary <string, float>();
if (catValue > dogValue)
{
result.Add("cat", catValue);
}
else
{
result.Add("dog", dogValue);
}
ClassificationCompleted?.Invoke(this, new ClassificationEventArgs(result));
}
catch (Exception ex)
{
Debug.WriteLine($"error: {ex.Message}");
}
}
}
catch (Exception e)
{
Console.WriteLine(e);
}
}