Struct CutOutAugmenter 

pub struct CutOutAugmenter {
    pub cutout_size: usize,
    pub probability: f64,
    pub fill_value: f64,
}

CutOut augmentation.

Randomly erases a fixed-size square region in the input data. Simpler variant of Random Erasing with fixed square regions.

Reference: DeVries & Taylor, “Improved Regularization of Convolutional Neural Networks with Cutout” (2017)

Parameters

• size: Size of the square region to erase
• p: Probability of applying cutout (default: 1.0)
• fill_value: Value to fill the erased region (default: 0.0)

Fields

cutout_size: usize

Size of the square cutout region.

probability: f64

Probability of applying cutout.

fill_value: f64

Fill value for erased region.

Implementations

impl CutOutAugmenter

pub fn new( cutout_size: usize, probability: f64, fill_value: f64, ) -> TrainResult<Self>

Create a new CutOut augmenter.

pub fn with_size(size: usize) -> TrainResult<Self>

Create with default parameters.

Trait Implementations

impl Clone for CutOutAugmenter

fn clone(&self) -> CutOutAugmenter

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl DataAugmenter for CutOutAugmenter

fn augment( &self, data: &ArrayView2<'_, f64>, rng: &mut StdRng, ) -> TrainResult<Array2<f64>>

Augment the given data.

impl Debug for CutOutAugmenter

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.