PadToSquare<T>

Models & Types · AiDotNet.Augmentation.Image

Pads an image to make it square while preserving the original content centered.

Inheritance: AugmentationBase<T, ImageTensor<T>> → ImageAugmenterBase<T> → PadToSquare<T>

For Beginners

If your image is 200x300 pixels (tall rectangle), this will add 50 pixels of padding to the left and right, making it 300x300 (square). The original image stays in the center.

How It Works

PadToSquare adds padding to the shorter dimension of an image so that height equals width. The original image is centered within the square, with padding distributed evenly on both sides. This is useful when models require square inputs but you want to preserve aspect ratio.

When to use:

  • When your model requires square input but images have varying aspect ratios
  • When you want to avoid distortion from non-uniform resizing
  • Object detection where distortion could change bounding box proportions

Constructors

Constructor Summary
PadToSquare(PaddingMode mode, double fillValue, double probability) Creates a new PadToSquare augmentation.

Properties

Property Summary
PaddingMode Mode { get; } Gets the padding mode.
double FillValue { get; } Gets the constant fill value (used when Mode is Constant).

Methods

Method Summary
IDictionary<string, object> GetParameters()