nvidia.dali.fn.element_extract(__input, /, *, element_map, bytes_per_sample_hint=[0], preserve=False, seed=-1, device=None, name=None)

Extracts one or more elements from input sequence.

The outputs are slices in the first (outermost) dimension of the input. There are as many outputs as the elements provided in the element_map.

For example, for element_map = [2, 0, 3] there will be three outputs, containing 2nd, 0th and 3rd element of the input sequences respectively.

The input layout, if provided, must begin with F dimension. The outputs will have one less dimension than the input, that is for FHWC inputs, the outputs will be HWC elements.

This operator expects sequence inputs.

Supported backends
  • ‘cpu’

  • ‘gpu’


__input (TensorList) – Input to the operator.

Keyword Arguments:
  • element_map (int or list of int) – Indices of the elements to extract.

  • bytes_per_sample_hint (int or list of int, optional, default = [0]) –

    Output size hint, in bytes per sample.

    If specified, the operator’s outputs residing in GPU or page-locked host memory will be preallocated to accommodate a batch of samples of this size.

  • preserve (bool, optional, default = False) – Prevents the operator from being removed from the graph even if its outputs are not used.

  • seed (int, optional, default = -1) –

    Random seed.

    If not provided, it will be populated based on the global seed of the pipeline.