The OpenVINO Analytics Plugin is an open-source plugin for Nx Meta to demonstrate person detection and tracking. It showcases person detection and tracking capabilities using Intel's OpenVINO toolkit.

This plugin serves as a reference implementation to help you build custom video analytics integration for Nx Meta. It is an example project and was never intended for production deployment.

System requirements

The OpenVINO plugin runs only on Intel processors. Before you begin, review the System Requirements for the Intel® Distribution of OpenVINO™ Toolkit.

Supported features

The plugin includes the following capabilities out of the box:

• Person detection: Identifies human targets within a video frame.

• Line crossing detection: Triggers an event when an object crosses a defined virtual boundary.

• Loitering detection: Tracks objects that remain in a specified area for a set duration.

• Motion detection: Highlights pixel motion within selected zones.

Use the source code

You can use the open-source repository in two ways:

• Build a custom plugin: Clone the repository and replace the default training models with your own custom Intel OpenVINO models.

• Use as a reference architecture: Inspect the codebase to understand how to structure your own analytics integration from scratch.

Build and install the plugin

Follow these steps to set up the plugin project locally:

1. Clone the repository to your local machine:

   git clone https://github.com/networkoptix/nx_open_source.git

2. Navigate to the plugin directory and review the build instructions in readme.md.

3. Open readme.md in the Nx Meta Metadata SDK for general guidelines on building and installing plugins on the Nx Witness Server.

Related resources

• How to Create a Video Analytics Plugin: Learn the core concepts of building custom analytics plugins.

• An Introduction to Plugin Debugging: Troubleshoot runtime issues, memory leaks, and communication errors.