Releases: dlstreamer/pipeline-zoo
Release 2023.0
Intel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Zoo
The Intel® DL Streamer Pipeline Zoo is a catalog of media and media analytics pipelines optimized for Intel® hardware. It includes tools for downloading pipelines and their dependencies and tools for measuring their performance.
Pipelines are organized according to the task they perform (what types of input they accept and what types of output they generate). Tasks and pipelines are defined in a platform and framework independent way to allow implementations in a variety of frameworks and for multiple platform targets.
IMPORTANT:
The Intel® DL Streamer Pipeline Zoo is provided as a set of tools for system evaluation and benchmarking and is not intended for deployment into production environments without modification.
The project is pre-production and under active development. Please expect breaking changes and use tagged versions for stable functionality.
For the details of supported platforms, please refer to System Requirements section.
New in this Release
| Title | High-level description |
|---|---|
| Alignment with Intel® DL Streamer Pipeline Framework 2023.0 | Pipeline Zoo now uses the 2023.0 image of Intel® DL Streamer Pipeline Framework as its base image |
| Compatibility with OpenVINO™ Toolkit 2023.0 | Pipeline Zoo has been updated to use the 2023.0 version of the OpenVINO™ Toolkit |
| New Models Added | For detailed information see the Pipeline Zoo Models Repository Pre-Release v0.0.9 |
| New Media Added | For more information see Pipeline Zoo Media Repository Pre-Release v0.0.10 |
Changed in this Release
Full Changelog: v0.0.7...v0.0.9
System Requirements
Please refer to Intel® DL Streamer documentation.
Legal Information
No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document.
Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.
This document contains information on products, services and/or processes in development. All information provided here is subject to change without notice. Contact your Intel representative to obtain the latest forecast, schedule, specifications and roadmaps.
The products and services described may contain defects or errors which may cause deviations from published specifications. Current characterized errata are available on request.
Intel and the Intel logo are trademarks of Intel Corporation in the U.S. and/or other countries.
*Other names and brands may be claimed as the property of others.
© 2023 Intel Corporation.
Release 2022.1
Intel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Zoo
The Intel® DL Streamer Pipeline Zoo is a catalog of media and media analytics pipelines optimized for Intel® hardware. It includes tools for downloading pipelines and their dependencies and tools for measuring their performance.
Pipelines are organized according to the task they perform (what types of input they accept and what types of output they generate). Tasks and pipelines are defined in a platform and framework independent way to allow implementations in a variety of frameworks and for multiple platform targets.
IMPORTANT:
The Intel® DL Streamer Pipeline Zoo is provided as a set of tools for system evaluation and benchmarking and is not intended for deployment into production environments without modification.
The project is pre-production and under active development. Please expect breaking changes and use tagged versions for stable functionality.
For the details of supported platforms, please refer to System Requirements section.
New in this Release
| Title | High-level description |
|---|---|
| Alignment with Intel® DL Streamer Pipeline Framework 2022.1 | Pipeline Zoo now uses the 2022.1 image of Intel® DL Streamer Pipeline Framework as its base image |
| Compatibility with OpenVINO™ Toolkit 2022.1 | Pipeline Zoo has been updated to use the 2022.1 version of the OpenVINO™ Toolkit |
| New models added | New object detection and object classification pipelines were added. These are based on the following models:
* efficient-b0 * ssdlite-mobilenet-v2 |
| Platform support updates | Pipeline Zoo has added full support for Alder Lake and Tiger Lake platforms |
| Improved Benchmarking | Time to compute stream density on high density cores was significantly reduced |
Changed in this Release
- Naming aligned with Intel® DL Streamer product version
Full Changelog: v0.0.6...v0.0.7
System Requirements
Please refer to Intel® DL Streamer documentation.
Legal Information
No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document.
Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.
This document contains information on products, services and/or processes in development. All information provided here is subject to change without notice. Contact your Intel representative to obtain the latest forecast, schedule, specifications and roadmaps.
The products and services described may contain defects or errors which may cause deviations from published specifications. Current characterized errata are available on request.
Intel and the Intel logo are trademarks of Intel Corporation in the U.S. and/or other countries.
*Other names and brands may be claimed as the property of others.
© 2022 Intel Corporation.
Release v0.0.6
Intel(R) Deep Learning Streamer Pipeline Zoo
The DL Streamer Pipeline Zoo is a catalog of media and media analytics pipelines optimized for Intel hardware. It includes tools for downloading pipelines and their dependencies and tools for measuring their performace.
Pipelines are organized according to the task they perform (what types of input they accept and what types of output they generate). Tasks and pipelines are defined in a platform and framework independent way to allow implementations in a variety of frameworks and for multiple platform targets.
IMPORTANT:
The DL Streamer Pipeline Zoo is provided as a set of tools for system evaluation and benchmarking and is not intended for deployment into production environments without modification.
The project is pre-production and under active development. Please expect breaking changes and use tagged versions for stable functionality.
Features Include:
| Simple command line interface | A single entrypoint for downloading and running media analytics pipelines along with media and model dependencies |
| DL Streamer Pipeline Runner | Pipeline implementations and optimizations using the DL Streamer Pipeline Framework |
| Platform specific settings | Pipeline runner settings tuned for optimal performance on different platform types (e.g. core, xeon) |
| Measurement Settings | Settings for measuring different scenarios including single stream throughput and stream density. Settings can be customized and saved for reuse. |
| Containerized environment | Dockerfiles, build and run scripts for launching a reproducable environment |
Release v0.0.6
This release contains minor bug fixes and enhancements:
- duration expands number of frames in media beyond 60 seconds if given (calculated at 30 fps)
- added dog_bark media for object classification
What's Changed
New Contributors
Full Changelog: v0.0.5...v0.0.6
Contributors
Assets 2
Initial Preview Release (v0.0.5)
Intel(R) Deep Learning Streamer Pipeline Zoo
The DL Streamer Pipeline Zoo is a catalog of media and media analytics pipelines optimized for Intel hardware. It includes tools for downloading pipelines and their dependencies and tools for measuring their performace.
Pipelines are organized according to the task they perform (what types of input they accept and what types of output they generate). Tasks and pipelines are defined in a platform and framework independent way to allow implementations in a variety of frameworks and for multiple platform targets.
IMPORTANT:
The DL Streamer Pipeline Zoo is provided as a set of tools for system evaluation and benchmarking and is not intended for deployment into production environments without modification.
The project is pre-production and under active development. Please expect breaking changes and use tagged versions for stable functionality.
Features Include:
| Simple command line interface | A single entrypoint for downloading and running media analytics pipelines along with media and model dependencies |
| DL Streamer Pipeline Runner | Pipeline implementations and optimizations using the DL Streamer Pipeline Framework |
| Platform specific settings | Pipeline runner settings tuned for optimal performance on different platform types (e.g. core, xeon) |
| Measurement Settings | Settings for measuring different scenarios including single stream throughput and stream density. Settings can be customized and saved for reuse. |
| Containerized environment | Dockerfiles, build and run scripts for launching a reproducable environment |
Initial Preview Release (v0.0.5)
The initial release contains support for the following tasks and pipelines using a DL Streamer Pipeline Runner:
- Object Detection
- od-h264-ssd-mobilenet-v1-coco
- od-h265-ssd-mobilenet-v1-coco
- Object Classification
- oc-h264-full-frame-resnet-50-tf
- oc-h265-full-frame-resnet-50-tf
- oc-h264-ssd-mobilenet-v1-coco-resnet-50-tf
-oc-h265-ssd-mobilenet-v1-coco-resnet-50-tf
- Decode VPP
- decode-h265
- decode-h264-bgra
And provides settings tuned for performance on:
- Xeon: Intel(R) Xeon(R) Gold 6336Y CPU @ 2.40GHz.
- Core: 11th Gen Intel(R) Core(TM) i7-1185G7 @ 3.00GH