public static Conv2DInfo computePool2DInfo(int[] inShape, int[] filterSize,
int[] strides, PadType pad, roundingMode roundingMode,
ConvDataFormat dataFormat = ConvDataFormat.channelsLast, Nullable <int> padValue = null)
{
var filterHeight = filterSize[0];
var filterWidth = filterSize[1];
int[] filterShape;
if (dataFormat == ConvDataFormat.channelsLast)
{
filterShape = new int[] { filterHeight, filterWidth, inShape[3], inShape[3] };
}
else if (dataFormat == ConvDataFormat.channelsFirst)
{
filterShape = new int[] { filterHeight, filterWidth, inShape[1], inShape[1] };
}
else
{
throw new Exception("Unknown dataFormat");
}
var dilations = 1;
return(computeConv2DInfo(
inShape, filterShape, strides, new int[] { dilations }, pad, roundingMode, false,
dataFormat, padValue));
}
public static Conv2DInfo computeConv2DInfo(int[] inShape, int[] filterShape,
int[] strides, int[] dilations, PadType pad,
roundingMode roundingMode = roundingMode.none, bool depthwise = false,
ConvDataFormat dataFormat = ConvDataFormat.channelsLast, Nullable <int> padValue = null)
{
var batchSize = -1;
var inHeight = -1;
var inWidth = -1;
var inChannels = -1;
if (dataFormat == ConvDataFormat.channelsLast)
{
batchSize = inShape[0];
inHeight = inShape[1];
inWidth = inShape[2];
inChannels = inShape[3];
}
else
{
batchSize = inShape[0];
inChannels = inShape[1];
inHeight = inShape[2];
inWidth = inShape[3];
}
var filterHeight = filterShape[0];
var filterWidth = filterShape[1];
var filterChannels = filterShape[3];
var strideHeight = strides[0];
var strideWidth = strides[1];
var dilationHeight = dilations[0];
int dilationWidth = 0;
if (dilations.Length > 1)
{
dilationWidth = dilations[1];
}
else
{
dilationWidth = dilations[0];
}
var effectiveFilterHeight = getEffectiveFilterSize(filterHeight, dilationHeight);
var effectiveFilterWidth =
getEffectiveFilterSize(filterWidth, dilationWidth);
var d = getPadAndOutInfo(
pad, inHeight, inWidth, strideHeight, strideWidth, effectiveFilterHeight,
effectiveFilterWidth, roundingMode, padValue);
var padInfo = d.Item1;
var outHeight = d.Item2;
var outWidth = d.Item3;
var outChannels = depthwise ? filterChannels * inChannels : filterChannels;
int[] outShape = null;
if (dataFormat == ConvDataFormat.channelsFirst)
{
outShape = new int[] { batchSize, outChannels, outHeight, outWidth };
}
else if (dataFormat == ConvDataFormat.channelsLast)
{
outShape = new int[] { batchSize, outHeight, outWidth, outChannels };
}
return(new Conv2DInfo()
{
batchSize = batchSize,
dataFormat = dataFormat,
inHeight = inHeight,
inWidth = inWidth,
inChannels = inChannels,
outHeight = outHeight,
outWidth = outWidth,
outChannels = outChannels,
padInfo = padInfo,
strideHeight = strideHeight,
strideWidth = strideWidth,
filterHeight = filterHeight,
filterWidth = filterWidth,
dilationHeight = dilationHeight,
dilationWidth = dilationWidth,
inShape = inShape,
outShape = outShape,
filterShape = filterShape
});
}
/// <summary>
/// Add a bias to a tensor.
/// </summary>
/// <param name="x">The tensor to add the bias to.</param>
/// <param name="bias">The bias to add to `x`. Must be 1D or the same rank as `x`.</param>
/// <param name="dataFormat"></param>
/// <returns> Result of the bias adding.</returns>
public static Tensor biasAdd(this Tensor x, Tensor bias, ConvDataFormat dataFormat = ConvDataFormat.channelsLast)
{
if (bias.Rank != 1 && bias.Rank != x.Rank)
{
throw new Exception(
"Unexpected bias dimensions: " + bias.Rank +
"; expected it to be 1 or " + x.Rank);
}
var biasShape = bias.Shape;
Tensor y = null;
if (x.Rank == 5)
{
if (dataFormat == ConvDataFormat.channelsFirst)
{
if (biasShape.Length == 1)
{
y = x.add(bias.reshape(shape(1, biasShape[0], 1, 1, 1)));
}
else
{
y = x.add(bias.reshape(
shape(1, biasShape[3], biasShape[0], biasShape[1], biasShape[2])));
}
}
else if (dataFormat == ConvDataFormat.channelsLast)
{
if (biasShape.Length == 1)
{
y = x.add(bias.reshape(shape(1, 1, 1, 1, biasShape[0])));
}
else
{
var nb = new List <int>();
nb.Add(1);
nb.AddRange(biasShape);
y = x.add(bias.reshape(nb.ToArray()));
}
}
}
else if (x.Rank == 4)
{
if (dataFormat == ConvDataFormat.channelsFirst)
{
if (biasShape.Length == 1)
{
y = x.add(bias.reshape(shape(1, biasShape[0], 1, 1)));
}
else
{
y = x.add(
bias.reshape(shape(1, biasShape[2], biasShape[0], biasShape[1])));
}
}
else if (dataFormat == ConvDataFormat.channelsLast)
{
if (biasShape.Length == 1)
{
y = x.add(bias.reshape(shape(1, 1, 1, biasShape[0])));
}
else
{
var nb = new List <int>();
nb.Add(1);
nb.AddRange(biasShape);
y = x.add(bias.reshape(nb.ToArray()));
}
}
}
else if (x.Rank == 3)
{
if (dataFormat == ConvDataFormat.channelsFirst)
{
if (biasShape.Length == 1)
{
y = x.add(bias.reshape(shape(1, biasShape[0], 1)));
}
else
{
y = x.add(bias.reshape(shape(1, biasShape[1], biasShape[0])));
}
}
else if (dataFormat == ConvDataFormat.channelsLast)
{
if (biasShape.Length == 1)
{
y = x.add(bias.reshape(shape(1, 1, biasShape[0])));
}
else
{
var nb = new List <int>();
nb.Add(1);
nb.AddRange(biasShape);
y = x.add(bias.reshape(nb.ToArray()));
}
}
}
else if (x.Rank < 3)
{
y = x.add(bias);
}
else
{
throw new Exception("Unsupported input Rank by biasAdd: " + x.Rank.ToString());
}
return(y);
}
