openvino linux

### OpenVINO on Linux Installation and Usage Guide

#### Prerequisites
Before installing the Intel® Distribution of OpenVINO™ toolkit, ensure that the system meets all prerequisites. For Ubuntu-based systems, it is essential to have a supported version such as Ubuntu 20.04 LTS or earlier versions like Ubuntu 18.04.3 LTS[^2]. The hardware should also be compatible with OpenVINO requirements.

#### Docker Image Setup for OpenVINO
For users preferring containerized environments, an updated Docker image from DockerHub can simplify setup significantly. Specifically targeting those using NVIDIA GPUs alongside CPUs within an Ubuntu environment:

docker pull openvino/ubuntu20_dev:latest
docker run -it --rm --net=host --name openvino openvino/ubuntu20_dev:latest

This command sequence pulls down the latest available OpenVINO development image built specifically for Ubuntu 20 and runs this image interactively without needing additional configuration due to automatic detection mechanisms provided by the official images[^1].

#### Installing Directly on Host Machine
Alternatively, direct installation onto the host machine involves downloading the installer package directly from the official website following detailed instructions at [official documentation](https://docs.openvinotoolkit.org/latest/_docs_install_guides_installing_openvino_linux.html):

- Downloading the appropriate offline/online installers.
- Executing the installation script after extracting files.
- Configuring environmental variables post-installation through `source /opt/intel/openvino/bin/setupvars.sh`.

#### Post-Installation Configuration
After successful installation, configuring the environment correctly ensures seamless integration between different components including TensorFlow*, Caffe*, ONNX models conversion utilities among others which are key features offered by Intel's proprietary edition aimed primarily towards inference acceleration across various platforms supporting Movidius Myriad X VPU devices too.

#### Verification Steps
To verify whether everything has been set up properly one could execute sample applications bundled inside `/opt/intel/openvino/deployment_tools/demo` directory where multiple pre-configured demos reside ready-to-use out-of-the-box showcasing capabilities ranging from object recognition over video streams to facial landmark estimation tasks etc.

--related questions--

1. What specific steps need attention when setting up GPU support during OpenVINO installations?
2. How does model optimizer tool work under OpenVINO framework converting third-party frameworks into IR format suitable for deployment?
3. Can you provide examples demonstrating how to deploy trained neural networks utilizing OpenVINO runtime APIs effectively?
4. Are there any differences in functionality between running OpenVINO via Docker compared to native setups?