C3dConvLayer Class

June 14, 2020 ยท View on GitHub

This class implements a layer that performs convolution on a set of three-dimensional multi-channel images. Zero-padding is supported.

Settings

Filters size

void SetFilterHeight( int filterHeight );
void SetFilterWidth( int filterWidth );
void SetFilterDepth( int filterDepth );
void SetFilterCount( int filterCount );

Sets the filters' size and number.

Convolution stride

void SetStrideHeight( int strideHeight );
void SetStrideWidth( int strideWidth );
void SetStrideDepth( int strideDepth );

Sets the convolution stride. By default, the stride is 1.

Padding

void SetPaddingHeight( int paddingHeight );
void SetPaddingWidth( int paddingWidth );
void SetPaddingDepth( int paddingDepth );

Sets the width, height, and depth of zero-padding that will be added around the image. For example, if you set the padding width to 1, two additional rectangles filled with zeros will be added to the image: one on the left and one on the right.

By default, no padding is used, and these values are equal to 0.

Using the free terms

void SetZeroFreeTerm(bool isZeroFreeTerm);

Specifies if the free terms should be used. If you set this value to true, the free terms vector will be set to all zeros and won't be trained. By default, this value is set to false.

Trainable parameters

Filters

CPtr<CDnnBlob> GetFilterData() const;

The filters are represented by a blob of the following dimensions:

  • BatchLength * BatchWidth * ListSize is equal to the number of filters used (GetFilterCount())
  • Height is equal to GetFilterHeight()
  • Width is equal to GetFilterWidth()
  • Depth is equal to GetFilterDepth()
  • Channels is equal to the inputs' Channels

Free terms

CPtr<CDnnBlob> GetFreeTermData() const;

The free terms are represented by a blob of the total size equal to the number of filters used (GetFilterCount()).

Inputs

Each input accepts a blob with several images. The dimensions of all inputs should be the same:

  • BatchLength * BatchWidth * ListSize - the number of images in the set.
  • Height - the images' height.
  • Width - the images' width.
  • Depth - the images' depth.
  • Channels - the number of channels the image format uses.

Outputs

For each input the layer has one output. It contains a blob with the result of the convolution. The output blob dimensions are:

  • BatchLength is equal to the input BatchLength.
  • BatchWidth is equal to the input BatchWidth.
  • ListSize is equal to the input ListSize.
  • Height can be calculated from the input Height as (2 * PaddingHeight + Height - (1 + DilationHeight * (FilterHeight - 1)))/StrideHeight + 1.
  • Width can be calculated from the input Width as (2 * PaddingWidth + Width - (1 + DilationWidth * (FilterWidth - 1)))/StrideWidth + 1.
  • Depth can be calculated from the input Depth as
    (2 * PaddingDepth + Depth - FilterDepth)/StrideDepth + 1.
  • Channels is equal to GetFilterCount().