Operators

Individual operator samples demonstrating specific CV-CUDA operations.

Overview

The operator samples show focused functionality for understanding specific operations:

  • Gaussian - Blur and smoothing with configurable kernel and sigma

  • Resize - Image resizing with various interpolation methods (linear, cubic, area, nearest)

  • Reformat - Tensor layout conversions (HWC, CHW, NHWC, NCHW)

  • Stack - Batch creation from multiple tensors for parallel processing

  • Label - Connected component labeling for region identification

These samples are perfect for:

  • Learning individual operator behavior

  • Understanding operator parameters

  • Quick experimentation

  • Building custom pipelines

Operator Samples

Common Usage Patterns

Single Operator

Simple, focused operation:

import cvcuda
from common import read_image, write_image

image = read_image("input.jpg")
result = cvcuda.gaussian(image, (5, 5), (1.0, 1.0))
write_image(result, "output.jpg")

Chaining Operators

Combine multiple operations:

image = read_image("input.jpg")
resized = cvcuda.resize(image, (224, 224, 3))
blurred = cvcuda.gaussian(resized, (5, 5), (1.0, 1.0))
write_image(blurred, "output.jpg")

Batch Processing

batch = cvcuda.stack([read_image(p) for p in paths])
processed = cvcuda.gaussian(batch, (5, 5), (1.0, 1.0))

See Also