BatchRandomColorJitter

class torchaug.batch_transforms.BatchRandomColorJitter(brightness=0, contrast=0, saturation=0, hue=0, p=0.5, num_rand_calls=-1, inplace=False, value_check=False)

Randomly change the brightness, contrast, saturation and hue to a batch of images.

The batch is expected to have [B, …, 1 or 3, H, W] shape, where … means an arbitrary number of dimensions.

Parameters:

• brightness (float | tuple[float, float], optional) – How much to jitter brightness. brightness factor is chosen uniformly from [max(0, 1 - brightness), 1 + brightness] or the given [min, max]. Should be non negative numbers.

  Default: 0

• contrast (float | tuple[float, float], optional) – How much to jitter contrast. contrast factor is chosen uniformly from [max(0, 1 - contrast), 1 + contrast] or the given [min, max]. Should be non-negative numbers.

  Default: 0

• saturation (float | tuple[float, float], optional) – How much to jitter saturation. saturation factor is chosen uniformly from [max(0, 1 - saturation), 1 + saturation] or the given [min, max]. Should be non negative numbers.

  Default: 0

• hue (float | tuple[float, float], optional) – How much to jitter hue. hue_factor is chosen uniformly from [-hue, hue] or the given [min, max]. Should have 0 <= hue <= 0.5 or -0.5 <= min <= max <= 0.5. To jitter hue, the pixel values of the input image has to be non-negative for conversion to HSV space; thus it does not work if you normalize your image to an interval with negative values, or use an interpolation that generates negative values before using this function.

  Default: 0

• p (float, optional) – Probability to apply color jitter.

  Default: 0.5

• num_rand_calls (int, optional) – Number of random calls performed to apply augmentations at different orders on sub-batches. If -1, B calls are performed. The maximum is 24 = 4!, adjusted automatically if num_rand_calls > 24.

  Default: -1

• inplace (bool, optional) – If True, perform inplace operation to save memory and time.

  Default: False

• value_check (bool, optional) – Bool to perform tensor value check. Might cause slow down on some devices because of synchronization or large batch size.

  Default: False

apply_transform(imgs)

Color jitter the batch of images.

Note

Apply different transformation orders based on num_rand_calls.

Parameters:

imgs (Tensor) – Batch of images to jitter.

Return type:

Tensor

Returns:

Randomly color jittered batch of images.

static get_params(brightness, contrast, saturation, hue, batch_size)

Get the parameters for the randomized transform to be applied on image.

Parameters:

• brightness (tensor (min, max), optional) – The range from which the brightness factor is chosen uniformly. Pass None to turn off the transformation.

• contrast (tensor (min, max), optional) – The range from which the contrast factor is chosen uniformly. Pass None to turn off the transformation.

• saturation (tensor (min, max), optional) – The range from which the saturation factor is chosen uniformly. Pass None to turn off the transformation.

• hue (Optional[Tensor]) – The range from which the hue_factor is chosen uniformly. Pass None to turn off the transformation.

• batch_size (int) – The number of samples to draw.

Return type:

tuple[Tensor, Optional[Tensor], Optional[Tensor], Optional[Tensor], Optional[Tensor]]