Release 2020.1

What’s New in This Release:

Title	High-level description
Object tracking on CPU (gvatrack)	A new GStreamer element for tracking objects with IOU (Intersection Over Union), zero-term and short-term algorithms. Allows to track objects between frames and reduces the need to run inference on each frame. It also assigns unique ID to each object
Python API and samples	Introduced Python package that provides access to inference results produced by GVA plugin for user applications using GStreamer Python API. Also, aligned C++ API to match Python API.
Processing custom Python code (gvapython)	A new GStreamer element that allows user defined Python functions to process a frame. e.g. face blurring, metadata conversion, feature vector processing for re-identification models etc.
Reshape support in gvadetect, gvaclassify and gvainference	A new property to change model input resolution for already converted IR format model
YoloV3 support in gvadetect	Extended gvadetect element to support YoloV3 output format conversion into bounding boxes
Developer Experience improvements	The following new documentation is added: GStreamer Video Analytics Usage Tutorial: A tutorial on how to create single or multiple classification pipelines with video files or web camera as input source. GStreamer Video Analytics Advanced Usage Tutorial: A tutorial on GVA advance usage, such as publishing inference results to JSON file, and converting CNN Models to Intermediate Representation. Custom-Processing: An overview and examples of options available to implement custom processing logic for pipeline buffers data, including GVA inference results. API reference: C++ and Python APIs reference, which also contains introduction to working with GVA inference results on API user level, and Python application example, which takes advantage of GVA Python API to do inference results custom post processing.
Performance Optimization	Added I420 color format to gvadetect, gvainfer and gvaclassify capabilities

Optimization notice:

For optimal performance on Intel® Xeon® Scalable Processors, use "pre-proc=ie" option in inference elements. Please note that this is currently supported only with open source version of OpenVINO™ Toolkit.

Release details including known issues and system requirements are in available in the attached Release Notes.

Getting started is available on the Wiki.