This is preview of the product functionality that is being introduced to gain early developer feedback. Comments, questions, and suggestions are encouraged and should be submitted to the GitHub* Issues page
This repository contains a collection of GStreamer* elements to enable CNN model based video analytics capabilities (such as object detection, classification, recognition) in GStreamer* framework. The complete solution leverages
- Open source GStreamer* framework for pipeline management
- GStreamer* plugins for input and output such as media files and real-time streaming from camera or network
- Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins based on VAAPI
and additionally installs the following Deep Learning specific elements from this repository
- Inference plugins leveraging Intel® OpenVINO™ Toolkit for high performance inference using CNN models
- Visualization of computer vision results (such as bounding boxes and labels of detected objects) on top of video stream
GStreamer* Video Analytics Plugins are licensed under MIT license.
- Refer to Intel® OpenVINO™ Toolkit for hardware requirements for inference elements;
- On platforms with Intel® Processor Graphics, refer to gstreamer-vaapi for HW accelerated video decode and encode requirements.
- Linux* system with kernel >= 4.15
- GStreamer framework >= 1.14
This link provides detailed instructions how to build plugins and run samples in docker container or directly on host machine.
See command-line examples and C++ example
Bugs and requests can be reported on Issues page
More details about pipeline construction and data flow between pipeline elements
More details about metadata generated by inference plugins and attached to video frames
More details how to prepare Tensorflow/Caffe and other models for usage in the inference plugins
Elements list and properties list per each element
If you have bug fix or an idea to improve the project, please first let us know and submit proposal description to Issues page as at this stage of the project pull requests not monitored.
* Other names and brands may be claimed as the property of others.