Export an ONNX model from a TorchVision classification model. Args: model: PyTorch model to export. output: Path to the outp

Export RetinaNet model with backbone + head + EfficientNMS (no complex decoding needed). Args: model: PyTorch RetinaNet model to exp

Export an ONNX model from a TorchVision segmentation model. Args: model: PyTorch model to export. output: Path to the output

Finalizes the performance benchmark data dictionary and saves it in the output folder as a JSON file. The benchmark data will be all

Create from __cuda_array_interface__ or object that implements it.

Returns the managed handle. * * @return The managed handle. */

Prepares list of tuples : op-class-name (str) and class for all the operators that can be benchmarked.

Prepares and returns an argparse command line argument parser for the scripts that supports auto-benchmarking. This parser guarantees that al

Returns a dictionary with keys being the variable names initialized exclusively during the setup call # and values being their values

* @brief Checks if the tensor layout contains a channel. * @return true if there's a channel, false otherwise. */

Convert Image to Tensor with zero-copy wrapping.

* @brief Creates a sub-layout consisting of the last n dimensions. * @param n The number of dimensions to include. * @return A new TensorLay

PyNvVideoCodec <-> CV-CUDA interoperability example.

CUDA Python <-> CV-CUDA interoperability example.

CuPy <-> CV-CUDA interoperability example.

CUDA Python <-> CV-CUDA interoperability example.

PyCUDA <-> CV-CUDA interoperability example.

Run all NumPy to GPU transfer examples.

NvImgCodec <-> CV-CUDA interoperability example.

PyTorch <-> CV-CUDA interoperability example.

Connected components labeling an image with CVCUDA.

Resize an image with CVCUDA.

Stack an image with CVCUDA.

Reformat an image with CVCUDA.

Gaussian an image with CVCUDA.

Classification with CVCUDA preprocessing and TensorRT inference.

Hello world CV-CUDA application.

Object detection with CVCUDA preprocessing and TensorRT+EfficientNMS inference.

Semantic segmentation with CVCUDA preprocessing and TensorRT inference.

The main function. This will run the comparison function to compare two benchmarking runs.

The main function. This will run the comparison function to compare two benchmarking runs.

Maximizes the GPU clocks. Useful to do it before any type of performance benchmarking. :param device_id: The GPU device ID whose clocks s

Merges the CPU and GPU NVTX range information dictionaries into one such that it contains all the information. The keys will be flat expressi