Compare DALI pipeline outputs with ndd outputs. Performs comprehensive comparison including: - Metadata validation (batch size, shapes,

If there is any work scheduled in the pipeline but not yet consumed

Encode a box list relative to an anchor collection. Args: boxes: BoxList holding N boxes to be encoded anchors: BoxList o

Evaluates the underlying lazy expression, if any. If the tensor is a result of a lazy evaluation, calling `evaluate` will cause the

* @brief Fill a range of bits with given value */

* @brief Encodes a segmentation mask in a way that is easy to access to * know how many foreground pixels are there and get the flat index *

@brief Round down and convert to integer

* @brief Obtains the stored handle * * The value is valid as long as the owning unique handle object is not destroyed, reset * or overwritten

Run the pipeline without returning the resulting buffers. If the pipeline was created with `exec_pipelined` option set to `True`, thi

Get the depth of the stack. Note, that by default there is at least one element - the global scope.

* Performs a test on a single test file. * @params test_file_idx @see TestFileDescriptor */

* @brief true if the argument is empty or the particular sample * has a 0-volume shape */

Removes a ready task with the highest priorty or waits for one to appear or * for a shutdown notification. */

Generate a Signature for given schema. Parameters ---------- schema : OpSchema Schema for the operator. api : str "fn

Wrapper for asserting warnings, optionally with pattern matching. Can be used as context manager or with callable: with assert_warns

AST representation.

Checks the outputs of the pipeline. `outputs` return value from pipeline `run` `ref_out` a batch or tuple of batches `ref

Copy contents of DALI tensor to PyTorch's Tensor. Parameters ---------- dali_tensor : nvidia.dali.backend.TensorCPU or nvidia.dali.b

The identity block is the block that has no conv layer at shortcut. Args: input_tensor: input tensor kernel_size: default 3, the kernel size

The number of dimensions of the samples in the batch. The "batch dimension" is not included - e.g. a batch of HWC is still a 3D obje

Power implementation. \param base value to exponentiate \param exp power to expontiate to \return function value stored in single-preicision

* @brief Reserves at least new_num_bytes of storage. * * @param new_num_bytes The minimum size, in bytes of the buffer. Note that alignment or

Record one result for thread_id (append).

Return the already filled windows

* @name Configure contiguity of allocations * @{ */ * @brief If the batch is backed by contiguous buffer */