tfm.vision.preprocess_ops.color_jitter

Stay organized with collections Save and categorize content based on your preferences.

Applies color jitter to an image, similarly to torchvision`s ColorJitter.

tfm.vision.preprocess_ops.color_jitter(
    image: tf.Tensor,
    brightness: Optional[float] = 0.0,
    contrast: Optional[float] = 0.0,
    saturation: Optional[float] = 0.0,
    seed: Optional[int] = None
) -> tf.Tensor

image (tf.Tensor): Of shape [height, width, 3] and type uint8. brightness (float, optional): Magnitude for brightness jitter. Defaults to 0. contrast (float, optional): Magnitude for contrast jitter. Defaults to 0. saturation (float, optional): Magnitude for saturation jitter. Defaults to 0. seed (int, optional): Random seed. Defaults to None.

tf.Tensor The augmented image of type uint8.