GridDistortion<T>

Models & Types · AiDotNet.Augmentation.Image

Applies grid-based distortion to an image.

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

For Beginners

Imagine overlaying a flexible grid on your image, then randomly nudging each grid intersection point. The image warps smoothly between the displaced points, creating natural-looking distortions.

How It Works

Grid distortion divides the image into a grid of cells and randomly displaces the grid intersection points, then interpolates the displacement across each cell. This creates smooth, locally varying distortions.

When to use:

  • OCR and document recognition
  • Medical image augmentation
  • Any task requiring smooth local distortions

Constructors

Constructor Summary
GridDistortion(int numSteps, double distortLimit, double fillValue, double probability) Creates a new grid distortion augmentation.

Properties

Property Summary
double DistortLimit { get; } Gets the maximum distortion magnitude.
double FillValue { get; } Gets the fill value for out-of-bounds pixels.
int NumSteps { get; } Gets the number of grid steps in each dimension.

Methods

Method Summary
IDictionary<string, object> GetParameters()