The OpenVINO™ Optimized Image with ONNX RT allows high-performance deep learning inference workloads deployed on Intel® architecture. Paired together, developers can deploy ONNX models on any Intel® hardware that drives cost, power and development efficiency. With cloud-to-edge flow validated, developers can deploy the cloud/pre-trained AI models as well as apps at the edge to solve industry use cases. Thus, bridging the gap between deploying cloud-developed AI solutions and edge devices such as Intel® CPUs, GPUs, VPUs, and FPGAs. OpenVINO™ not only provides the heterogeneous hardware flexibility as single inference engine for deep learning but also other options such as Deep Learning Workbench and Reference Implementations.

Benefits:

• High-performance, fully tested and optimized AI Container
• Deploy anywhere – In the Cloud, On-Prem or at the Edge
• Pre-validated with ONNX Model Zoo
• Works with Azure ML and other Azure Services
• Use it to improve Scoring latency & Efficiency of models with Azure Services

To pull the image, use docker pull with the base image and tags below.

Base Image: mcr.microsoft.com/azureml/onnxruntime

Default Tags:

• CPU - :latest-openvino-cpu
• GPU - :latest-openvino-gpu
• MYRIAD - :latest-openvino-myriad
• VAD-M - :latest-openvino-vadm

Note: While the above is default, developers can also dynamically switch the target hardware. Learn more about building the image from Dockerfile here.

Get started right away on Intel® hardware for free using Intel® DevCloud. Along with several examples, developers can get started with the Clean Room Worker Safety Jupyter notebook using a Tiny Yolo V2 ONNX model for object detection. Developers can also acquire developer kits from partners to jump start with hardware and software tools to prototype, test and deploy a solution. Learn more about the kits here.

By downloading and using this container and the included software, you agree to the terms and conditions of the Intel® License Agreement.