DALI Release 0.29.0

The DALI 0.29.0 is not yet a major release, so the features, functionality, and performance might be limited.

Using DALI 0.29.0

To upgrade to DALI 0.29.0 from an older version of DALI, follow the installation and usage information in the DALI User Guide.

Note: The internal DALI C++ API used for operator’s implementation, and the C++ API that enables using DALI as a library from native code, is not yet officially supported. Hence these APIs may change in the next release without advance notice.

Key Features and Enhancements

This DALI release includes the following key features and enhancements.

  • New operators:

    • Added the NumpyReader GPU Operator based on GPUDirect Storage (#2477).

    • CUDA-acceleared JPEG2000 image decoding was enabled in the ImageDecoder operator that uses the nvJpeg2k library(#2501).

    • Added the segmentation.RandomMaskPixel operator to create random masks that contain foreground pixels (#2445).

    • Added OneHot for GPUs (#2436).

  • Moved all NVTX infrastructure into core and create the DALI domain (#2472).

  • New Examples:

    • Added mask processing to the COCO reader with an Augmentations example (#2426).

    • Added a reductions example (#2457).

    • An example of random_mask_pixel was updated to perform biased random crop (#2474).

    • Updated the ExternalSource framework examples (#2482).

  • Operator Improvements:

    • Pad: Added support for the per-sample shape and alignment requirements (#2432)

    • RandomResizedCrop: Enabled channel-first and video support + add tests (#2430)

    • PythonFunction operator: Added support for output layouts (#2486).

    • Optimized the DCT GPU kernel (#2471).

    • COCOReader: Added support for uncompressed RLE masks (#2478).

    • Improved transforms.Rotation to accept scalar inputs (#2494).

  • DALI now supports CUDA 11.1 update 1 (#2419).

Fixed Issues

This DALI release includes the following fixes.

  • NumpyReader: To fix ABI incompatibility issues, replaced the std::regex with custom implementation (#2489) .

  • Fixed the dimensionality of labels in SSDRandomCrop (#2488)

Breaking Changes

  • Python 3.5 is no longer supported by the official DALI wheels.

Deprecated Features

Deprecate the squeeze_labels option from MXNet iterator and enhanced the .squeeze function to match the numpy-style interface (#2450)

Known Issues

  • The video loader operator requires that the key frames occur, at a minimum, every 10 to 15 frames of the video stream.

    If the key frames occur at a frequency that is less than 10-15 frames, the returned frames might be out of sync.

  • The DALI TensorFlow plugin might not be compatible with TensorFlow versions 1.15.0 and later.

    To use DALI with the TensorFlow version that does not have a prebuilt plugin binary that is shipped with DALI, make sure that the compiler that is used to build TensorFlow exists on the system during the plugin installation. (Depending on the particular version, you can use GCC 4.8.4, GCC 4.8.5, or GCC 5.4.)

  • Due to some known issues with meltdown/spectra mitigations and DALI, DALI shows the best performance when running in Docker with escalated privileges, for example:
    • privileged=yes in Extra Settings for AWS data points
    • --privileged or --security-opt seccomp=unconfined for bare Docker