NEW Intel® Distribution of OpenVINO™ toolkit 2021.4 Long-Term Support (LTS) RELEASE

[JesusE_Intel]

Update to the latest Long-Term Support (LTS) release (recommended for developers taking solutions into production) 

• This new 2021.4 Long-Term Support (LTS) Release provides bug fixes, two years of support and maintenance ensuring application stability while reducing the risks and costs associated with more frequent version upgrades. 
• New Jupyter Notebooks, demos and support for additional public models to make development easier: 
  • Ready-to-run Jupyter Notebooks with tutorials for converting TensorFlow and PyTorch models, image classification, segmentation, depth estimation, post-training quantization and more. (https://github.com/openvinotoolkit/openvino_notebooks) 
  • Audio Noise Suppression & Time Series Forecasting demos. (https://github.com/openvinotoolkit/open_model_zoo/tree/master/demos) 
  • Public Models: RCAN and IseeBetter (image super-resolution), Attention OCR (image text prediction), Tacotron 2 (text-to-speech) and ModNet (portrait/image matting)
• Significant performance improvements on time-to-first-inference latency for both CPU and integrated GPU (iGPU), as a result of doing more initialization work in parallel among other optimizations.
• FPGA Support is Ending: Intel® is transitioning to the next-generation programmable deep learning solution, Intel® FPGA AI Suite and will support OpenVINO™ toolkit when productized. 
• 2020.3.2 LTS was the final release to include support for Intel® Vision Accelerator Design with an Intel® Arria® 10 FPGA and the Intel® Programmable Acceleration Card with Intel® Arria® 10 GX FPGA.​  
• Customer inquiries should be directed to your Intel® Programmable Solutions Group account manager or subscribe to get notified with the latest updates.
• Public preview of new open-source component called OpenVINO™ integration with TensorFlow designed to enable TensorFlow developers to try OpenVINO with minimal code changes. (https://github.com/openvinotoolkit/openvino_tensorflow) 
• Note: Adopting native OpenVINO API is recommended for highest level of performance, lowest memory footprint and complete hardware control.

RESOURCES 

• LTS Release Notes 
• Long-Term Support (LTS) Policy 
• Get Started 
• Documentation  
• Learning Resources