Class: AdditiveGaussianNoiseAugmenter

AdditiveGaussianNoiseAugmenter(sigma)

Adds noise sampled from a gaussian distribution

Warning : This function is slow, use ia.additiveNoise with tensorflowjs for fast noise generation

Constructor

new AdditiveGaussianNoiseAugmenter(sigma)

Parameters:
Name Type Description
sigma Object | Number | Hasard

options, if number, this is sigma

Properties
Name Type Attributes Default Description
mean Number <optional>
0

mean of the gaussian distribution

sigma Number

sigma of the gaussian distribution

scale Number <optional>
1

if scale is defined (0 < scale < 1), then the noise can apply at a less granular scale

perChannel Number <optional>
false

If perChannel is true, then the sampled values may be different per channel (and pixel).

Source:
Examples
// Simple usage
ia.additiveGaussianNoise(3);
//Simple usage with random variable
ia.additiveGaussianNoise(h.number(0, 3));
// Explicit usage with random variable
ia.additiveGaussianNoise({
	sigma: 3,
	scale: 0.5,
	perChannel: true,
	mean: 5
});
// Explicit usage with random variable
ia.additiveGaussianNoise({
	sigma: h.number(0, 3),
	scale: h.number(0.2, 1),
	perChannel: h.boolean(),
	mean: h.number(-5, 5)
});