protected override void Write(RiffWriter writer, Mat input)
{
var dataDepth = input.Depth == Depth.U8 ? Depth.U8 : Depth.S16;
var elementSize = input.Depth == Depth.U8 ? 1 : 2;
var step = elementSize * input.Cols;
var data = new byte[step * input.Rows];
var dataHandle = GCHandle.Alloc(data, GCHandleType.Pinned);
try
{
var dataAddr = dataHandle.AddrOfPinnedObject();
using (var dataHeader = new Mat(input.Cols, input.Rows, dataDepth, input.Channels,
dataAddr, elementSize * input.Rows))
{
if (input.Depth != dataDepth)
{
using (var conversion = new Mat(input.Size, dataDepth, input.Channels))
{
CV.Convert(input, conversion);
CV.Transpose(conversion, dataHeader);
}
}
else
{
CV.Transpose(input, dataHeader);
}
}
}
finally { dataHandle.Free(); }
writer.Write(data);
}
public override IObservable <TArray> Process <TArray>(IObservable <TArray> source)
{
var outputFactory = ArrFactory <TArray> .TemplateFactory;
var accumulatorFactory = ArrFactory <TArray> .TemplateSizeChannelFactory;
return(Observable.Defer(() =>
{
TArray accumulator = null;
return source.Select(input =>
{
if (accumulator == null)
{
accumulator = accumulatorFactory(input, Depth.F32);
CV.Convert(input, accumulator);
return input;
}
else
{
var output = outputFactory(input);
CV.RunningAvg(input, accumulator, Alpha);
CV.Convert(accumulator, output);
return output;
}
});
}));
}
Mat Calibrate(Tuple <Mat, Mat> source)
{
// Do we need to check array sizes or let CV take care of that?
//if (source.Item1.Size != source.Item2.Size)
if (!calibration_set)
{
// Read the calibration files
offset_raw = ReadFully(CreateStream(OffsetPath));
slope_raw = ReadFully(CreateStream(SlopePath));
calibration_set = true;
}
// Use exposures to index into raw data
var exposures = source.Item2;
var offset_dat = new double[source.Item1.Rows * source.Item1.Cols];
var slope_dat = new double[source.Item1.Rows * source.Item1.Cols];
// Fast copy
unsafe
{
fixed(byte *p_offset_raw = offset_raw)
fixed(double *p_offset_dat = offset_dat)
fixed(byte *p_slope_raw = slope_raw)
fixed(double *p_slope_dat = slope_dat)
{
double *p_offset_raw_d = (double *)p_offset_raw;
double *p_slope_raw_d = (double *)p_slope_raw;
var numel = exposures.Rows * exposures.Cols;
for (int i = 0; i < exposures.Rows; i++)
{
for (int j = 0; j < exposures.Cols; j++)
{
var mat_off = (int)exposures[i, j].Val0 * numel;
offset_dat[exposures.Cols * i + j] = p_offset_raw_d[mat_off + exposures.Cols * i + j];
slope_dat[exposures.Cols * i + j] = p_slope_raw_d[mat_off + exposures.Cols * i + j];
}
}
}
}
// Generate Mats from data
var offset = Mat.CreateMatHeader(offset_dat, source.Item1.Rows, source.Item1.Cols, Depth.F64, 1);
var slope = Mat.CreateMatHeader(slope_dat, source.Item1.Rows, source.Item1.Cols, Depth.F64, 1);
// Calibrate
if (source.Item1.Depth != Depth.F64)
{
var tmp = new Mat(source.Item1.Size, Depth.F64, 1);
CV.Convert(source.Item1, tmp);
return(tmp * (slope - offset));
}
else
{
return(source.Item1 * (slope - offset));
}
}
static IObservable <double[]> AutoThreshold <TOther>(IObservable <IObservable <object> > source, IObservable <TOther> bufferBoundaries, Func <double> scale)
{
var concat = new Concat {
Axis = 1
};
var firstBoundary = Observable.Return <TOther>(default(TOther));
bufferBoundaries = firstBoundary.Concat(bufferBoundaries);
return(bufferBoundaries.Publish(ps => ps.Window(2).Skip(1).SelectMany(start => Observable.Defer(() =>
{
var n = 0;
var mean = default(double[]);
var variance = default(double[]);
var buffer = default(double[]);
return source.SelectMany(xs => xs.Cast <Mat>()).TakeUntil(ps).Select(xs =>
{
if (mean == null)
{
mean = new double[xs.Rows];
variance = new double[xs.Rows];
buffer = new double[xs.Rows * xs.Cols];
}
// Convert data into temporary buffer
var bufferHandle = GCHandle.Alloc(buffer, GCHandleType.Pinned);
try
{
using (var bufferHeader = new Mat(xs.Rows, xs.Cols, Depth.F64, 1, bufferHandle.AddrOfPinnedObject()))
{
CV.Convert(xs, bufferHeader);
}
}
finally { bufferHandle.Free(); }
// Knuth's online variance algorithm
// http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance
for (int i = 0; i < buffer.Length; i++)
{
var row = i / xs.Cols;
var col = i % xs.Cols;
var previousMean = mean[row];
var delta = buffer[i] - previousMean;
var newMean = previousMean + delta / (n + i + 1);
variance[row] = variance[row] + delta * (buffer[i] - newMean);
mean[row] = newMean;
}
n += xs.Cols;
return xs;
}).TakeLast(1).Select(xs =>
{
var result = new double[xs.Rows];
for (int i = 0; i < result.Length; i++)
{
result[i] = mean[i] + scale() * Math.Sqrt(variance[i] / (n - 1));
}
return result;
});
}))));
}
public static void UpdateBuffer(int bid, Mat buffer, int sampleRate)
{
var transpose = buffer.Rows < buffer.Cols;
var channels = transpose ? buffer.Rows : buffer.Cols;
if (channels > 2)
{
throw new InvalidOperationException("Unsupported number of channels for the specified output format.");
}
var format = channels == 2 ? ALFormat.Stereo16 : ALFormat.Mono16;
var convertDepth = buffer.Depth != Depth.S16;
if (convertDepth || transpose)
{
// Convert if needed
if (convertDepth)
{
var temp = new Mat(buffer.Rows, buffer.Cols, Depth.S16, 1);
CV.Convert(buffer, temp);
buffer = temp;
}
// Transpose multichannel to column order
if (transpose)
{
var temp = new Mat(buffer.Cols, buffer.Rows, buffer.Depth, 1);
CV.Transpose(buffer, temp);
buffer = temp;
}
}
AL.BufferData(bid, format, buffer.Data, buffer.Rows * buffer.Step, sampleRate);
}
public override IObservable <Mat> Process(IObservable <Mat> source)
{
return(Observable.Using(
() => PulsePalManager.ReserveConnection(PortName),
pulsePal => source.Do(input =>
{
var pulseInterval = 1000000 / (Frequency * CycleTimeMicroseconds);
var pulseTimes = new int[input.Cols];
var pulseVoltages = new byte[input.Cols];
for (int i = 0; i < pulseTimes.Length; i++)
{
pulseTimes[i] = pulseInterval * i;
}
using (var voltageHeader = Mat.CreateMatHeader(pulseVoltages))
{
CV.Convert(input, voltageHeader);
}
lock (pulsePal.PulsePal)
{
pulsePal.PulsePal.SendCustomPulseTrain(PulseId, pulseTimes, pulseVoltages);
}
})));
}
private Mat GetEphysDataS16(short[,] data)
{
var output = new Mat(NumberOfEphysChannels, NumberOfSamples, Depth.S16, 1);
using (var header = Mat.CreateMatHeader(data))
{
CV.Convert(header, output);
}
return(output);
}
private Mat GetData(float[,] data)
{
var output = new Mat(NumberOfChannels, NumberOfSamples, Depth.F32, 1);
using (var header = Mat.CreateMatHeader(data))
{
CV.Convert(header, output);
}
return(output);
}
private Mat GetDelta(double[] data)
{
var output = new Mat(1, NumberOfFrames, Depth.F64, 1);
using (var header = Mat.CreateMatHeader(data))
{
CV.Convert(header, output);
}
return(output);
}
static TElement[] Process <TArray, TElement>(TArray input, Depth depth)
where TArray : Arr
where TElement : struct
{
var inputHeader = input.GetMat();
var output = new TElement[inputHeader.Rows * inputHeader.Cols];
using (var outputHeader = Mat.CreateMatHeader(output, inputHeader.Rows, inputHeader.Cols, depth, 1))
{
CV.Convert(inputHeader, outputHeader);
}
return(output);
}
private static Mat GetEphysData(ushort[,] data)
{
var numChannels = data.GetLength(0);
var numSamples = data.GetLength(1);
var output = new Mat(numChannels, numSamples, Depth.U16, 1);
using (var header = Mat.CreateMatHeader(data))
{
CV.Convert(header, output);
}
return(output);
}
public override IObservable <IplImage> Process(IObservable <IplImage> source)
{
return(source.Select(input =>
{
if (input.Depth == IplDepth.U16)
{
var temp = new IplImage(input.Size, IplDepth.F32, input.Channels);
CV.Convert(input, temp);
input = temp;
}
var output = new IplImage(input.Size, IplDepth.U8, input.Channels);
CV.Threshold(input, output, ThresholdValue, MaxValue, ThresholdType);
return output;
}));
}
Mat GetState(int[] data)
{
if (data.Length == 0)
{
return(null);
}
var output = new Mat(1, data.Length, Depth.S32, 1);
using (var header = Mat.CreateMatHeader(data, 1, data.Length, Depth.S32, 1))
{
CV.Convert(header, output);
}
return(output);
}
/// <summary>
/// Converts the given value object to the specified type, using the specified
/// context and culture information.
/// </summary>
/// <param name="context">
/// An <see cref="ITypeDescriptorContext"/> that provides a format context.
/// </param>
/// <param name="culture">
/// A <see cref="CultureInfo"/>. If <b>null</b> is passed, the current culture
/// is assumed.
/// </param>
/// <param name="value">The <see cref="Object"/> to convert.</param>
/// <param name="destinationType">The <see cref="Type"/> to convert the value parameter to.</param>
/// <returns>An <see cref="Object"/> that represents the converted value.</returns>
public override object ConvertTo(ITypeDescriptorContext context, CultureInfo culture, object value, Type destinationType)
{
if (value != null && destinationType == typeof(string))
{
var mat = (Mat)value;
var array = new float[mat.Rows, mat.Cols];
using (var arrayHeader = Mat.CreateMatHeader(array))
{
CV.Convert(mat, arrayHeader);
}
return(ArrayConvert.ToString(array, culture));
}
return(base.ConvertTo(context, culture, value, destinationType));
}
public override IObservable <Mat> Process(IObservable <Mat> source)
{
return(source.Select(input =>
{
var channels = Channels;
var output = new Mat(input.Size, input.Depth, input.Channels);
var reference = new Mat(1, input.Cols, input.Depth, input.Channels);
if (channels == null || channels.Length == 0)
{
if (input.Depth != Depth.F32)
{
var temp = new Mat(reference.Rows, reference.Cols, Depth.F32, reference.Channels);
CV.Reduce(input, temp, 0, ReduceOperation.Avg);
CV.Convert(temp, reference);
}
else
{
CV.Reduce(input, reference, 0, ReduceOperation.Avg);
}
}
else if (channels.Length == 1)
{
CV.Copy(input.GetRow(channels[0]), reference);
}
else
{
var sum = input.Depth != Depth.F32
? new Mat(reference.Rows, reference.Cols, Depth.F32, reference.Channels)
: reference;
sum.SetZero();
for (int i = 0; i < channels.Length; i++)
{
using (var referenceChannel = input.GetRow(channels[i]))
{
CV.Add(sum, referenceChannel, sum);
}
}
CV.ConvertScale(sum, reference, 1f / channels.Length);
}
CV.Repeat(reference, output);
CV.Sub(input, output, output);
return output;
}));
}
Mat GetAuxiliaryData(int[,] data)
{
if (data.Length == 0)
{
return(null);
}
var numChannels = data.GetLength(0);
var numSamples = data.GetLength(1);
var output = new Mat(numChannels, numSamples, Depth.U16, 1);
using (var header = Mat.CreateMatHeader(data))
{
CV.Convert(header, output);
}
return(output);
}
private Mat GetAuxiliaryData(ushort[,] data)
{
using (var header = Mat.CreateMatHeader(data))
{
if (AuxFormat == RHD2164Configuration.AuxDataFormat.Volts)
{
var output = new Mat(NumberOfAuxChannels, NumberOfSamples, Depth.F32, 1);
CV.ConvertScale(header, output, 0.0000374);
return(output);
}
else
{
var output = new Mat(NumberOfAuxChannels, NumberOfSamples, Depth.U16, 1);
CV.Convert(header, output);
return(output);
}
}
}
public override IObservable <TArray> Process <TArray>(IObservable <TArray> source)
{
var outputFactory = ArrFactory <TArray> .TemplateSizeFactory;
var inputFactory = ArrFactory <TArray> .TemplateSizeChannelFactory;
return(source.Select(input =>
{
var output = outputFactory(input, Depth.F32, 2);
if (input.ElementType != output.ElementType)
{
var temp = inputFactory(input, Depth.F32);
CV.Convert(input, temp);
input = temp;
}
CV.DFT(input, output, OperationFlags, 0);
return output;
}));
}
public GenericDevice()
{
// Reference to context
this.oni_ref = ONIManager.ReserveDAQ();
source = Observable.Create <Mat>(observer =>
{
EventHandler <FrameReceivedEventArgs> inputReceived;
oni_ref.Environment.Start();
inputReceived = (sender, e) =>
{
var frame = e.Value;
// If this frame contains data from the selected device_index
if (frame.DeviceIndices.Contains(DeviceIndex))
{
var dat = frame.Data <ushort>(DeviceIndex, ReadSize);
var mat = new Mat(1, dat.Length, ElementDepth, 1);
using (var header = Mat.CreateMatHeader(dat))
{
CV.Convert(header, mat);
}
observer.OnNext(mat);
}
};
oni_ref.Environment.FrameInputReceived += inputReceived;
return(Disposable.Create(() =>
{
oni_ref.Environment.FrameInputReceived -= inputReceived;
oni_ref.Dispose();
}));
});
}
// Mat case
public override IObservable <Mat> Process(IObservable <Mat> source)
{
// TODO: what happens if more than one frame needs to be processed befor this finishes?
return(source.Do(
input =>
{
// Sanity check
if (rows < input.Rows || cols < input.Cols)
{
throw new IndexOutOfRangeException();
}
// Data to send (row indicator along with input exposure pattern)
var data = new Mat(rows, cols, Depth.S32, 1); //S32
var sub_data = data.GetSubRect(new Rect(1, 0, input.Cols, input.Rows));
// Convert element type if needed
var convertDepth = input.Depth != Depth.S32; //S32
if (convertDepth)
{
CV.Convert(input, sub_data);
}
else
{
CV.Copy(input, sub_data);
}
// Write out matrix, row by row with the first number being an encoded row number
for (int i = 0; i < rows; i++)
{
var row = data.GetRow(i);
row[0] = new Scalar(i + 16384, 0, 0, 0);
oni_ref.DAQ.Write((uint)DeviceIndex.SelectedIndex, row.Data, 4 * (data.Cols));
}
}));
}
public override IObservable <Mat> Process(IObservable <Mat> source)
{
return(Observable.Defer(() =>
{
int rows = 0;
double[] data = null;
double[] feedforwardCoefficients = null;
double[] feedbackCoefficients = null;
double[] dataWeights = null;
double[] dataMemory = null;
double[] outputWeights = null;
double[] outputMemory = null;
return source.Select(input =>
{
if (FeedforwardCoefficients != feedforwardCoefficients ||
FeedbackCoefficients != feedbackCoefficients ||
rows != input.Rows ||
data != null && data.Length != rows * input.Cols)
{
rows = input.Rows;
feedforwardCoefficients = FeedforwardCoefficients;
feedbackCoefficients = FeedbackCoefficients;
dataWeights = InitializeWeights(feedforwardCoefficients);
outputWeights = InitializeWeights(feedbackCoefficients);
for (int i = 0; i < outputWeights.Length - 1; i++)
{
outputWeights[i] = -outputWeights[i];
}
if (dataWeights != IdentityWeight || outputWeights != IdentityWeight)
{
data = new double[rows * input.Cols];
dataMemory = new double[rows * (dataWeights.Length - 1)];
outputMemory = new double[rows * (outputWeights.Length - 1)];
}
}
if (dataWeights == IdentityWeight && outputWeights == IdentityWeight)
{
return input;
}
else
{
var dataHandle = GCHandle.Alloc(data, GCHandleType.Pinned);
try
{
var output = new Mat(input.Size, input.Depth, input.Channels);
using (var dataHeader = new Mat(input.Size, Depth.F64, 1, dataHandle.AddrOfPinnedObject()))
{
CV.Convert(input, dataHeader);
ProcessData(rows, data, dataWeights, dataMemory, outputWeights, outputMemory);
CV.Convert(dataHeader, output);
}
return output;
}
finally { dataHandle.Free(); }
}
});
}));
}
public override IObservable <Mat> Process(IObservable <Tuple <Mat, Mat> > source)
{
return(Observable.Create <Mat>(observer =>
{
bool active = false;
var activeBuffers = new List <SampleBuffer>();
return source.Subscribe(input =>
{
try
{
var data = input.Item1;
var trigger = input.Item2;
// Update pending windows
activeBuffers.RemoveAll(buffer =>
{
buffer.Update(data, 0);
if (buffer.Completed)
{
// Window is ready, emit
observer.OnNext(buffer.Samples);
return true;
}
return false;
});
// Check if new triggers have arrived
var nonZero = CV.CountNonZero(trigger);
if (nonZero <= 0)
{
active = false;
}
else
{
var triggerBuffer = new byte[trigger.Cols];
var triggerHandle = GCHandle.Alloc(triggerBuffer, GCHandleType.Pinned);
using (var triggerHeader = new Mat(1, triggerBuffer.Length, Depth.U8, 1, triggerHandle.AddrOfPinnedObject()))
{
CV.Convert(trigger, triggerHeader);
triggerHandle.Free();
}
for (int i = 0; i < triggerBuffer.Length; i++)
{
var triggerHigh = triggerBuffer[i] > 0;
if (triggerHigh && !active)
{
var buffer = new SampleBuffer(data, Count, i);
buffer.Update(data, i);
if (buffer.Completed)
{
// Window is ready, emit
observer.OnNext(buffer.Samples);
}
// Window is missing data, add to list
else
{
activeBuffers.Add(buffer);
}
}
active = triggerHigh;
}
}
}
catch (Exception ex)
{
observer.OnError(ex);
}
},
error => observer.OnError(error),
() => observer.OnCompleted());
}));
}
public override IObservable <Pose> Process(IObservable <IplImage> source)
{
return(Observable.Defer(() =>
{
TFSessionOptions options = new TFSessionOptions();
unsafe
{
byte[] GPUConfig = new byte[] { 0x32, 0x02, 0x20, 0x01 };
fixed(void *ptr = &GPUConfig[0])
{
options.SetConfig(new IntPtr(ptr), GPUConfig.Length);
}
}
var graph = new TFGraph();
var session = new TFSession(graph, options, null);
var bytes = File.ReadAllBytes(ModelFileName);
graph.Import(bytes);
TFTensor tensor = null;
var config = ConfigHelper.PoseConfig(PoseConfigFileName);
return source.Select(input =>
{
if (tensor == null || tensor.GetTensorDimension(1) != input.Height || tensor.GetTensorDimension(2) != input.Width)
{
tensor = new TFTensor(
TFDataType.Float,
new long[] { 1, input.Height, input.Width, 3 },
input.WidthStep * input.Height * 4);
}
using (var image = new IplImage(input.Size, IplDepth.F32, 3, tensor.Data))
{
CV.Convert(input, image);
}
var runner = session.GetRunner();
runner.AddInput(graph["Placeholder"][0], tensor);
runner.Fetch(graph["concat_1"][0]);
// Run the model
var output = runner.Run();
// Fetch the results from output:
var poseTensor = output[0];
var pose = new Mat((int)poseTensor.Shape[0], (int)poseTensor.Shape[1], Depth.F32, 1, poseTensor.Data);
var result = new Pose(input);
var threshold = MinConfidence;
for (int i = 0; i < pose.Rows; i++)
{
BodyPart bodyPart;
bodyPart.Name = config[i];
bodyPart.Confidence = (float)pose.GetReal(i, 2);
if (bodyPart.Confidence < threshold)
{
bodyPart.Position = new Point2f(float.NaN, float.NaN);
}
else
{
bodyPart.Position.X = (float)pose.GetReal(i, 1);
bodyPart.Position.Y = (float)pose.GetReal(i, 0);
}
result.Add(bodyPart);
}
return result;
});
}));
}
public override IObservable <IplImage> Process(IObservable <IplImage> source)
{
return(Observable.Defer(() =>
{
int averageCount = 0;
IplImage image = null;
IplImage difference = null;
IplImage background = null;
return source.Select(input =>
{
if (background == null || background.Size != input.Size)
{
averageCount = 0;
image = new IplImage(input.Size, IplDepth.F32, input.Channels);
difference = new IplImage(input.Size, IplDepth.F32, input.Channels);
background = new IplImage(input.Size, IplDepth.F32, input.Channels);
background.SetZero();
}
var output = new IplImage(input.Size, IplDepth.U8, input.Channels);
if (averageCount < BackgroundFrames)
{
averageCount++;
output.SetZero();
CV.Acc(input, background);
if (averageCount == BackgroundFrames)
{
CV.ConvertScale(background, background, 1.0 / averageCount, 0);
}
}
else
{
CV.Convert(input, image);
switch (SubtractionMethod)
{
case SubtractionMethod.Bright:
CV.Sub(image, background, difference);
break;
case SubtractionMethod.Dark:
CV.Sub(background, image, difference);
break;
case SubtractionMethod.Absolute:
default:
CV.AbsDiff(image, background, difference);
break;
}
if (AdaptationRate > 0)
{
CV.RunningAvg(image, background, AdaptationRate);
}
CV.Threshold(difference, output, ThresholdValue, 255, ThresholdType);
}
return output;
});
}));
}
public override IObservable <Pose> Process(IObservable <IplImage> source)
{
return(Observable.Defer(() =>
{
TFSessionOptions options = new TFSessionOptions();
unsafe
{
byte[] GPUConfig = new byte[] { 0x32, 0x02, 0x20, 0x01 };
fixed(void *ptr = &GPUConfig[0])
{
options.SetConfig(new IntPtr(ptr), GPUConfig.Length);
}
}
var graph = new TFGraph();
var session = new TFSession(graph, options, null);
var bytes = File.ReadAllBytes(ModelFileName);
graph.Import(bytes);
IplImage temp = null;
TFTensor tensor = null;
TFSession.Runner runner = null;
var config = ConfigHelper.PoseConfig(PoseConfigFileName);
return source.Select(input =>
{
var poseScale = 1.0;
const int TensorChannels = 3;
var frameSize = input.Size;
var scaleFactor = ScaleFactor;
if (scaleFactor.HasValue)
{
poseScale = scaleFactor.Value;
frameSize.Width = (int)(frameSize.Width * poseScale);
frameSize.Height = (int)(frameSize.Height * poseScale);
poseScale = 1.0 / poseScale;
}
if (tensor == null || tensor.GetTensorDimension(1) != frameSize.Height || tensor.GetTensorDimension(2) != frameSize.Width)
{
tensor = new TFTensor(
TFDataType.Float,
new long[] { 1, frameSize.Height, frameSize.Width, TensorChannels },
frameSize.Width * frameSize.Height * TensorChannels * sizeof(float));
runner = session.GetRunner();
runner.AddInput(graph["Placeholder"][0], tensor);
runner.Fetch(graph["concat_1"][0]);
}
var frame = input;
if (frameSize != input.Size)
{
if (temp == null || temp.Size != frameSize)
{
temp = new IplImage(frameSize, input.Depth, input.Channels);
}
CV.Resize(input, temp);
frame = temp;
}
using (var image = new IplImage(frameSize, IplDepth.F32, TensorChannels, tensor.Data))
{
CV.Convert(frame, image);
}
// Run the model
var output = runner.Run();
// Fetch the results from output:
var poseTensor = output[0];
var pose = new Mat((int)poseTensor.Shape[0], (int)poseTensor.Shape[1], Depth.F32, 1, poseTensor.Data);
var result = new Pose(input);
var threshold = MinConfidence;
for (int i = 0; i < pose.Rows; i++)
{
BodyPart bodyPart;
bodyPart.Name = config[i];
bodyPart.Confidence = (float)pose.GetReal(i, 2);
if (bodyPart.Confidence < threshold)
{
bodyPart.Position = new Point2f(float.NaN, float.NaN);
}
else
{
bodyPart.Position.X = (float)(pose.GetReal(i, 1) * poseScale);
bodyPart.Position.Y = (float)(pose.GetReal(i, 0) * poseScale);
}
result.Add(bodyPart);
}
return result;
});
}));
}