public _Tensor[] QLearningTensors()
{
Texture2D[] textures = textureQueue.ToArray();
_Tensor[] tensorCube = new _Tensor[textures.Length];
float[] yChannelValues;
for (int i = 0; i < textures.Length; i++)
{
pixels = textures[i].GetPixels();
currentTextures[i] = new Texture2D(textures[i].width, textures[i].height, textureFormat, false);
currentTextures[i].SetPixels(Preprocessing.RGB2Gray(pixels));
currentTextures[i].Apply();
yChannelValues = new float[pixels.Length];
for (int j = 0; j < pixels.Length; j++)
{
yChannelValues[j] = Preprocessing.RGB2Gray(pixels[j]).a;
}
tensorCube[i] = new _Tensor(yChannelValues, textures[i].width, textures[i].height);
}
return(tensorCube);
}
public double[,,] GetInputTensor()
{
Texture2D[] textures = textureQueue.ToArray();
double[,,] result = new double[rtHeight, rtWidth, stepFrames];
int totalLength = rtHeight * rtWidth;
float[] yValues;
for (int i = 0; i < textures.Length; i++)
{
pixels = textures[i].GetPixels();
yValues = new float[pixels.Length];
for (int j = 0; j < yValues.Length; j++)
{
yValues[0] = Preprocessing.RGB2Gray(pixels[j]).a;
}
int row = 0;
int column = 0;
for (int pos = 0; pos < totalLength; pos++)
{
result[row, column, i] = yValues[pos];
column = (column + 1) % result.GetLength(1);
if (column == 0)
{
row++;
}
}
}
return(result);
}