Functional API

Quick start


This API is experimental and subject to change without notice!

Functional API is designed to simplify the usage of DALI operators in a psuedo-imperative way. It exposes operators as functions, with ths same name as the operator class, but converted to snake_case - for example ops.FileReader will be exposed as fn.file_reader().


import nvidia.dali as dali

pipe = dali.pipeline.Pipeline(batch_size = 3, num_threads = 2, device_id = 0)
with pipe:
    files, labels = dali.fn.file_reader(file_root = "./my_file_root")
    images = dali.fn.image_decoder(files, device = "mixed")
    images = dali.fn.rotate(images, angle = dali.fn.uniform(range=(-45,45)))
    images = dali.fn.resize(images, resize_x = 300, resize_y = 300)
    pipe.set_outputs(images, labels)
outputs =

The use of functional API does not change other aspects of pipeline definition - the functions still operate on and return pipeline.DataNode objects.

Interoperability with operator objects

Functional API is, for the major part, only a wrapper around operator objects - as such, it is inherently compatible with the object-based API. The following example mixes the two, using object API to pre-configure a file reader and a resize operator:

pipe = dali.pipeline.Pipeline(batch_size = 3, num_threads = 2, device_id = 0)
reader = dali.ops.FileReader(file_root = ".")
resize = dali.ops.Resize(device = "gpu", resize_x = 300, resize_y = 300)

with pipe:
    files, labels = reader()
    images = dali.fn.image_decoder(files, device = "mixed")
    images = dali.fn.rotate(images, angle = dali.fn.uniform(range=(-45,45)))
    images = resize(images)
    pipe.set_outputs(images, labels)
outputs =


nvidia.dali.fn.audio_decoder(*inputs, **arguments)

see nvidia.dali.ops.AudioDecoder

nvidia.dali.fn.batch_permutation(*inputs, **arguments)

see nvidia.dali.ops.BatchPermutation

nvidia.dali.fn.bb_flip(*inputs, **arguments)

see nvidia.dali.ops.BbFlip

nvidia.dali.fn.bbox_paste(*inputs, **arguments)

see nvidia.dali.ops.BBoxPaste

nvidia.dali.fn.box_encoder(*inputs, **arguments)

see nvidia.dali.ops.BoxEncoder

nvidia.dali.fn.brightness(*inputs, **arguments)

see nvidia.dali.ops.Brightness

nvidia.dali.fn.brightness_contrast(*inputs, **arguments)

see nvidia.dali.ops.BrightnessContrast

nvidia.dali.fn.caffe2_reader(*inputs, **arguments)

see nvidia.dali.ops.Caffe2Reader

nvidia.dali.fn.caffe_reader(*inputs, **arguments)

see nvidia.dali.ops.CaffeReader

nvidia.dali.fn.cast(*inputs, **arguments)

see nvidia.dali.ops.Cast*inputs, **arguments)

see nvidia.dali.ops.Cat

nvidia.dali.fn.coco_reader(*inputs, **arguments)

see nvidia.dali.ops.COCOReader

nvidia.dali.fn.coin_flip(*inputs, **arguments)

see nvidia.dali.ops.CoinFlip

nvidia.dali.fn.color_space_conversion(*inputs, **arguments)

see nvidia.dali.ops.ColorSpaceConversion

nvidia.dali.fn.color_twist(*inputs, **arguments)

see nvidia.dali.ops.ColorTwist

nvidia.dali.fn.constant(*inputs, **arguments)

see nvidia.dali.ops.Constant

nvidia.dali.fn.contrast(*inputs, **arguments)

see nvidia.dali.ops.Contrast

nvidia.dali.fn.coord_flip(*inputs, **arguments)

see nvidia.dali.ops.CoordFlip

nvidia.dali.fn.coord_transform(*inputs, **arguments)

see nvidia.dali.ops.CoordTransform

nvidia.dali.fn.copy(*inputs, **arguments)

see nvidia.dali.ops.Copy

nvidia.dali.fn.crop(*inputs, **arguments)

see nvidia.dali.ops.Crop

nvidia.dali.fn.crop_mirror_normalize(*inputs, **arguments)

see nvidia.dali.ops.CropMirrorNormalize

nvidia.dali.fn.dl_tensor_python_function(*inputs, **arguments)

see nvidia.dali.ops.DLTensorPythonFunction

nvidia.dali.fn.dump_image(*inputs, **arguments)

see nvidia.dali.ops.DumpImage

nvidia.dali.fn.element_extract(*inputs, **arguments)

see nvidia.dali.ops.ElementExtract

nvidia.dali.fn.erase(*inputs, **arguments)

see nvidia.dali.ops.Erase

nvidia.dali.fn.external_source(source=None, num_outputs=None, *, cycle=None, name=None, device='cpu', layout=None, cuda_stream=None, use_copy_kernel=None, batch=True, **kwargs)

Creates a data node which is populated with data from a Python source. The data can be provided by the source function or iterable, or it can be provided by pipeline.feed_input(name, data, layout, cuda_stream) inside pipeline.iter_setup.

In the case of the GPU input, it is the user responsibility to modify the provided GPU memory content only using provided stream (DALI schedules a copy on it and all work is properly queued). If no stream is provided feeding input blocks until the provided memory is copied to the internal buffer.


nvidia.dali.ops.ExternalSource() operator is not compatible with TensorFlow integration.


To return a batch of copies of the same tensor, use nvidia.dali.types.Constant(), which is more performant.

  • source (callable or iterable) –

    The source of the data.

    The source is polled for data (via a call source() or next(source)) when the pipeline needs input for the next iteration. Depending on the value of num_outputs, the source can supply one or more data items. The data item can be a whole batch (default) or a single batch entry (when batch==False). If num_outputs is not set, the source is expected to return one item (a batch or a sample). If this value is specified (even if its value is 1), the data is expected to a be tuple, or list, where each element corresponds to respective return value of the external_source.

    The data samples must be in one of the compatible array types:
    • NumPy ndarray (CPU)

    • MXNet ndarray (CPU)

    • PyTorch tensor (CPU or GPU)

    • CuPy array (GPU)

    • objects implementing __cuda_array_interface__

    • DALI Tensor object

    Batch sources must produce entire batches of data. This can be achieved either by adding a new outermost dimension to an array or by returning a list of arrays (in which case they can be of different size, but must have the same rank and element type). A batch source can also produce a DALI TensorList object, which can be an output of another DALI pipeline.

    A per-batch source may accept one positional argument. If it does, it is the index of current iteration within epoch and consecutive calls will be source(0), source(1), and so on.

    A per-sample source may accept one positional argument of type nvidia.dali.types.SampleInfo, which contains index of the sample in current epoch and in the batch, as well as current iteration number.

    If the source is a generator function, the function is invoked and treated as an iterable. However, unlike a generator, the function can be used with cycle. In this case, the function will be called again when the generator reaches the end of iteration.

    For GPU inputs, it is a user’s responsibility to modify the provided GPU memory content only in the provided stream. DALI schedules a copy on this stream, and all work is properly queued. If no stream is provided, DALI will use a default, with a best-effort approach at correctness. See the cuda_stream argument documentation for more information.

  • num_outputs (int, optional) –

    If specified, denotes the number of TensorLists that are produced by the source function.

    If set, the operator returns a list of DataNode objects, otherwise a single DataNode object is returned.

Keyword Arguments
  • cycle (bool, optional) –

    If set to True, the source will be wrapped.

    If set to False, StopIteration is raised when the end of data is reached. This flag requires that the source is a collection, for example, an iterable object where iter(source) returns a fresh iterator on each call or a gensource erator function. In the latter case, the generator function is called again when more data than was yielded by the function is requested.

  • name (str, optional) –

    The name of the data node.

    Used when feeding the data in iter_setup and can be omitted if the data is provided by source.

  • layout (layout str or list/tuple thereof, optional) –

    If provided, sets the layout of the data.

    When num_outputs > 1, the layout can be a list that contains a distinct layout for each output. If the list has fewer than num_outputs elements, only the first outputs have the layout set, the rest of the outputs don’t have a layout set.

  • cuda_stream (optional, cudaStream_t or an object convertible to cudaStream_t, such as cupy.cuda.Stream or torch.cuda.Stream) –

    The CUDA stream is used to copy data to the GPU or from a GPU source.

    If this parameter is not set, a best-effort will be taken to maintain correctness. That is, if the data is provided as a tensor/array from a recognized library such as CuPy or PyTorch, the library’s current stream is used. Although this approach works in typical scenarios, with advanced use cases, and code that uses unsupported libraries, you might need to explicitly supply the stream handle.

    This argument has two special values:
    • 0 - Use the default CUDA stream

    • 1 - Use DALI’s internal stream

    If internal stream is used, the call to feed_input will block until the copy to internal buffer is complete, since there’s no way to synchronize with this stream to prevent overwriting the array with new data in another stream.

  • use_copy_kernel (bool, optional) –

    If set to True, DALI will use a CUDA kernel to feed the data instead of cudaMemcpyAsync (default).


    This is applicable only when copying data to and from GPU memory.

  • blocking (bool, optional) – Determines whether the external source should wait until data is available or just fail when the data is not available.

  • no_copy (boo, optional) –

    Determines whether DALI should copy the buffer when feed_input is called.

    If set to True, DALI passes the user memory directly to the pipeline, instead of copying it. It is the user responsibility to keep the buffer alive and unmodified until it is consumed by the pipeline.

    The buffer can be modified or freed again after the output of the relevant iterations has been consumed. Effectively, it happens after prefetch_queue_depth or cpu_queue_depth * gpu_queue_depth (when they are not equal) iterations following the feed_input call.

    The memory location must match the specified device parameter of the operator. For the CPU, the provided memory can be one contiguous buffer or a list of contiguous Tensors. For the GPU, to avoid extra copy, the provided buffer must be contiguous. If you provide a list of separate Tensors, there will be an additional copy made internally, consuming both memory and bandwidth.

  • batch (bool, optional) – If set to True or None, the source is expected to produce an entire batch at once. If set to False, the source is called per-sample.

nvidia.dali.fn.fast_resize_crop_mirror(*inputs, **arguments)

see nvidia.dali.ops.FastResizeCropMirror

nvidia.dali.fn.file_reader(*inputs, **arguments)

see nvidia.dali.ops.FileReader

nvidia.dali.fn.flip(*inputs, **arguments)

see nvidia.dali.ops.Flip

nvidia.dali.fn.gaussian_blur(*inputs, **arguments)

see nvidia.dali.ops.GaussianBlur

nvidia.dali.fn.hsv(*inputs, **arguments)

see nvidia.dali.ops.Hsv

nvidia.dali.fn.hue(*inputs, **arguments)

see nvidia.dali.ops.Hue

nvidia.dali.fn.image_decoder(*inputs, **arguments)

see nvidia.dali.ops.ImageDecoder

nvidia.dali.fn.image_decoder_crop(*inputs, **arguments)

see nvidia.dali.ops.ImageDecoderCrop

nvidia.dali.fn.image_decoder_random_crop(*inputs, **arguments)

see nvidia.dali.ops.ImageDecoderRandomCrop

nvidia.dali.fn.image_decoder_slice(*inputs, **arguments)

see nvidia.dali.ops.ImageDecoderSlice

nvidia.dali.fn.jitter(*inputs, **arguments)

see nvidia.dali.ops.Jitter

nvidia.dali.fn.lookup_table(*inputs, **arguments)

see nvidia.dali.ops.LookupTable

nvidia.dali.fn.mel_filter_bank(*inputs, **arguments)

see nvidia.dali.ops.MelFilterBank

nvidia.dali.fn.mfcc(*inputs, **arguments)

see nvidia.dali.ops.MFCC

nvidia.dali.fn.mxnet_reader(*inputs, **arguments)

see nvidia.dali.ops.MXNetReader

nvidia.dali.fn.nemo_asr_reader(*inputs, **arguments)

see nvidia.dali.ops.NemoAsrReader

nvidia.dali.fn.nonsilent_region(*inputs, **arguments)

see nvidia.dali.ops.NonsilentRegion

nvidia.dali.fn.normal_distribution(*inputs, **arguments)

see nvidia.dali.ops.NormalDistribution

nvidia.dali.fn.normalize(*inputs, **arguments)

see nvidia.dali.ops.Normalize

nvidia.dali.fn.numpy_reader(*inputs, **arguments)

see nvidia.dali.ops.NumpyReader

nvidia.dali.fn.old_color_twist(*inputs, **arguments)

see nvidia.dali.ops.OldColorTwist

nvidia.dali.fn.one_hot(*inputs, **arguments)

see nvidia.dali.ops.OneHot

nvidia.dali.fn.optical_flow(*inputs, **arguments)

see nvidia.dali.ops.OpticalFlow

nvidia.dali.fn.pad(*inputs, **arguments)

see nvidia.dali.ops.Pad

nvidia.dali.fn.paste(*inputs, **arguments)

see nvidia.dali.ops.Paste

nvidia.dali.fn.peek_image_shape(*inputs, **arguments)

see nvidia.dali.ops.PeekImageShape

nvidia.dali.fn.permute_batch(*inputs, **arguments)

see nvidia.dali.ops.PermuteBatch

nvidia.dali.fn.power_spectrum(*inputs, **arguments)

see nvidia.dali.ops.PowerSpectrum

nvidia.dali.fn.preemphasis_filter(*inputs, **arguments)

see nvidia.dali.ops.PreemphasisFilter

nvidia.dali.fn.python_function(*inputs, **arguments)

see nvidia.dali.ops.PythonFunction

nvidia.dali.fn.random_bbox_crop(*inputs, **arguments)

see nvidia.dali.ops.RandomBBoxCrop

nvidia.dali.fn.random_resized_crop(*inputs, **arguments)

see nvidia.dali.ops.RandomResizedCrop

nvidia.dali.fn.reinterpret(*inputs, **arguments)

see nvidia.dali.ops.Reinterpret

nvidia.dali.fn.reshape(*inputs, **arguments)

see nvidia.dali.ops.Reshape

nvidia.dali.fn.resize(*inputs, **arguments)

see nvidia.dali.ops.Resize

nvidia.dali.fn.resize_crop_mirror(*inputs, **arguments)

see nvidia.dali.ops.ResizeCropMirror

nvidia.dali.fn.rotate(*inputs, **arguments)

see nvidia.dali.ops.Rotate

nvidia.dali.fn.saturation(*inputs, **arguments)

see nvidia.dali.ops.Saturation

nvidia.dali.fn.sequence_reader(*inputs, **arguments)

see nvidia.dali.ops.SequenceReader

nvidia.dali.fn.sequence_rearrange(*inputs, **arguments)

see nvidia.dali.ops.SequenceRearrange

nvidia.dali.fn.shapes(*inputs, **arguments)

see nvidia.dali.ops.Shapes

nvidia.dali.fn.slice(*inputs, **arguments)

see nvidia.dali.ops.Slice

nvidia.dali.fn.spectrogram(*inputs, **arguments)

see nvidia.dali.ops.Spectrogram

nvidia.dali.fn.sphere(*inputs, **arguments)

see nvidia.dali.ops.Sphere

nvidia.dali.fn.ssd_random_crop(*inputs, **arguments)

see nvidia.dali.ops.SSDRandomCrop

nvidia.dali.fn.stack(*inputs, **arguments)

see nvidia.dali.ops.Stack

nvidia.dali.fn.tfrecord_reader(*inputs, **arguments)

see nvidia.dali.ops.TFRecordReader

nvidia.dali.fn.to_decibels(*inputs, **arguments)

see nvidia.dali.ops.ToDecibels

nvidia.dali.fn.transpose(*inputs, **arguments)

see nvidia.dali.ops.Transpose

nvidia.dali.fn.uniform(*inputs, **arguments)

see nvidia.dali.ops.Uniform

nvidia.dali.fn.video_reader(*inputs, **arguments)

see nvidia.dali.ops.VideoReader

nvidia.dali.fn.video_reader_resize(*inputs, **arguments)

see nvidia.dali.ops.VideoReaderResize

nvidia.dali.fn.warp_affine(*inputs, **arguments)

see nvidia.dali.ops.WarpAffine

nvidia.dali.fn.water(*inputs, **arguments)

see nvidia.dali.ops.Water

nvidia.dali.fn.reductions.max(*inputs, **arguments)

see nvidia.dali.ops.reductions.Max

nvidia.dali.fn.reductions.mean(*inputs, **arguments)

see nvidia.dali.ops.reductions.Mean

nvidia.dali.fn.reductions.mean_square(*inputs, **arguments)

see nvidia.dali.ops.reductions.MeanSquare

nvidia.dali.fn.reductions.min(*inputs, **arguments)

see nvidia.dali.ops.reductions.Min

nvidia.dali.fn.reductions.rms(*inputs, **arguments)

see nvidia.dali.ops.reductions.RMS

nvidia.dali.fn.reductions.std_dev(*inputs, **arguments)

see nvidia.dali.ops.reductions.StdDev

nvidia.dali.fn.reductions.sum(*inputs, **arguments)

see nvidia.dali.ops.reductions.Sum

nvidia.dali.fn.reductions.variance(*inputs, **arguments)

see nvidia.dali.ops.reductions.Variance

nvidia.dali.fn.segmentation.random_mask_pixel(*inputs, **arguments)

see nvidia.dali.ops.segmentation.RandomMaskPixel

nvidia.dali.fn.segmentation.select_masks(*inputs, **arguments)

see nvidia.dali.ops.segmentation.SelectMasks

nvidia.dali.fn.transforms.combine(*inputs, **arguments)

see nvidia.dali.ops.transforms.Combine

nvidia.dali.fn.transforms.crop(*inputs, **arguments)

see nvidia.dali.ops.transforms.Crop

nvidia.dali.fn.transforms.rotation(*inputs, **arguments)

see nvidia.dali.ops.transforms.Rotation

nvidia.dali.fn.transforms.scale(*inputs, **arguments)

see nvidia.dali.ops.transforms.Scale

nvidia.dali.fn.transforms.shear(*inputs, **arguments)

see nvidia.dali.ops.transforms.Shear

nvidia.dali.fn.transforms.translation(*inputs, **arguments)

see nvidia.dali.ops.transforms.Translation