Adjust Contrast

group Adjust Contrast

Functions

NVCVStatus cvcudaAdjustContrastCreate(NVCVOperatorHandle *handle)

Constructs an instance of the AdjustContrast operator.

Parameters:

handle – [out] Where the image instance handle will be written to.

Must not be NULL.

Return values:

NVCV_ERROR_INVALID_ARGUMENT – Handle is null.
NVCV_ERROR_OUT_OF_MEMORY – Not enough memory to create the operator.
NVCV_SUCCESS – Operation executed successfully.

NVCVStatus cvcudaAdjustContrastSubmit(NVCVOperatorHandle handle, cudaStream_t stream, NVCVTensorHandle in, NVCVTensorHandle out, double contrastFactor)

Executes the AdjustContrast operation on the given cuda stream. This operation does not wait for completion.

Blends each image toward its grayscale mean by a single scalar factor, matching torchvision.transforms.v2.functional.adjust_contrast:

  out = clamp(contrastFactor * in + (1 - contrastFactor) * mean, 0, bound)

where bound is 1.0 for floating-point images and the dtype maximum for integer images, and mean is the per-image mean of its grayscale conversion. Grayscale uses the BT.601 luma weights 0.2989 R + 0.587 G + 0.114 B (floored for integer 3-channel input, matching torchvision); a single-channel image is its own grayscale. A factor of 0 produces a flat gray image, 1 leaves the image unchanged, and values > 1 increase contrast. Integer results round-to-nearest, so they may differ from torchvision (which truncates) by at most 1 LSB. Three-channel tensors are interpreted in RGB component order. Three-channel image batches use their RGB format semantics; formats without a color model are interpreted as RGB. The format swizzle is honored, including BGR storage order.

Reference: torchvision.transforms.v2.functional.adjust_contrast (out = clamp(factor * in + (1 - factor) * grayscale_mean, 0, bound)).

Limitations:

Input: Data Layout: [kNHWC, kHWC, kNCHW, kCHW] Channels: [1, 3]

Data Type	Allowed
8bit Unsigned	Yes
8bit Signed	No
16bit Unsigned	No
16bit Signed	No
32bit Unsigned	No
32bit Signed	No
16bit Float	No
32bit Float	Yes
64bit Float	No

Output: Data Layout: [kNHWC, kHWC, kNCHW, kCHW] Channels: [1, 3]

Data Type	Allowed
8bit Unsigned	Yes
8bit Signed	No
16bit Unsigned	No
16bit Signed	No
32bit Unsigned	No
32bit Signed	No
16bit Float	No
32bit Float	Yes
64bit Float	No

Input/Output dependency

Property	Input == Out
Data Layout	Yes
Data Type	Yes
Channels	Yes
Width	Yes
Height	Yes
Samples	Yes

Parameters:

handle – [in] Handle to the operator.
- Must not be NULL.
stream – [in] Handle to a valid CUDA stream.
in – [in] Input tensor to get values from.
- Must not be NULL.
out – [out] Output tensor to set values to.
- Must not be NULL.
contrastFactor – [in] Non-negative scalar contrast multiplier.

Return values:

NVCV_ERROR_INVALID_ARGUMENT – Some parameter is outside valid range.
NVCV_ERROR_INTERNAL – Internal error in the operator, invalid types passed in.
NVCV_SUCCESS – Operation executed successfully.

NVCVStatus cvcudaAdjustContrastVarShapeSubmit(NVCVOperatorHandle handle, cudaStream_t stream, NVCVImageBatchHandle in, NVCVImageBatchHandle out, double contrastFactor)

Executes the AdjustContrast operation on a batch of images.

Apart from input and output image batches, all parameters are the same as cvcudaAdjustContrastSubmit.

Parameters:

in – [in] Input image batch.
out – [out] Output image batch.