PixelShuffleLayer<T>

Layers · AiDotNet.NeuralNetworks.Layers

Pixel shuffle (sub-pixel convolution) layer for efficient spatial upsampling.

Inheritance: LayerBase<T> → PixelShuffleLayer<T>

Implements: IShapeContract

For Beginners

Imagine you have a small image and want to make it bigger.

Pixel shuffle works by:

  1. Starting with extra channel information (4x more channels for 2x upscaling)
  2. Rearranging those channel values into spatial positions
  3. Creating a larger image with the same amount of total information

For example, with 2x upscaling:

  • Input: 64 channels × 32×32 pixels
  • Output: 16 channels × 64×64 pixels (same total data, different arrangement)

This is commonly used in super-resolution models like Real-ESRGAN and ESPCN.

How It Works

Pixel shuffle rearranges elements from the channel dimension into spatial dimensions, effectively upscaling the spatial resolution. This is more efficient than transposed convolution (deconvolution) for upsampling operations.

For a 2x upscaling, the layer takes 4 channel values and arranges them as a 2x2 spatial block. The operation follows the formula: [batch, channels * r^2, height, width] -> [batch, channels, height * r, width * r]

Constructors

Constructor Summary
PixelShuffleLayer(int upscaleFactor) Initializes a new instance of the PixelShuffleLayer class.

Properties

Property Summary
bool SupportsTraining { get; }
int UpscaleFactor { get; } Gets the upscaling factor used by this layer.

Methods

Method Summary
IReadOnlyList<OutputAxisContract> OutputAxesFor(int inputRank)
void ResetState()