Skip to content

RandomNoise

RandomNoise class

dlf.preprocessing.noise.RandomNoise(mean=0.0, std=5.0, color=True)

Data augmentation method that randomly adds Gaussian noise to a RGB image

Aliases

  • noise
  • random_noise
  • Noise
  • RandomNoise

Arguments

  • mean: float. Mean value, centre of the distribution. Defaults to 0.
  • std: float. Standard deviation or width of the distribution. Defaults to 5.
  • color: bool. If true, noise is added to color channels separately. Defaults to True.

YAML Configuration

preprocess_list:
    noise:
        mean: 0.0
        std: 0.8
        color: False