Upsampling2DLayer.md

CUpsampling2DLayer Class

• CUpsampling2DLayer Class
  • Settings
    • Height multiplier
    • Width-multiplier
  • Trainable parameters
  • Inputs
  • Outputs

The class implements a layer that scales up a set of two-dimensional multi-channel images. The new pixels are filled up by repeating the existing pixels' values.

Settings

Height multiplier

void SetHeightCopyCount( int newHeightCopyCount );

Width multiplier

void SetWidthCopyCount( int newHeightCopyCount );

Trainable parameters

The layer has no trainable parameters.

Inputs

Each of the inputs accepts a blob of int or float data with a set of images:

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

Outputs

For each input, the corresponding output returns a blob with the upscaled image. Upsampling is performed by repetition of original pixels without any interpolation.

The blob dimensions are:

• BatchLength, BatchWidth, ListSize, Depth, and Channels equal the input dimensions
• Height is GetHeightCopyCount() times larger than the input Height
• Width is GetWidthCopyCount() times larger than the input Width