Introduction

Vision Development Module supports loading and executing third party Deep Learning framework models. The models from the following Deep Learning frameworks are supported.

  • TensorFlow Inference Engine
  • OpenVINO™ Inference Engine

The Deep Learning Inference Engines enable user to:

  • Load pre-trained Deep Learning Models into Software and Hardware ecosystem
  • Run loaded models in Windows and Real Time targets
  • Supply Vision Image and LabVIEW data to learned models

Supported Platforms

The following platforms are supported:

Development Environments

  • LabVIEW 64-bit

TensorFlow Runtime and Real Time Targets Support:

  • Windows 64-bit
  • NI Linux RT 64-bit

OpenVINO™ Runtime and Real Time Targets Support:

  • Windows 64-bit (Windows 7 Embedded Standard is not supported)
  • NI Linux RT 64-bit

When to Use

The Deep Learning Inference Engines provide the ability to load and execute third party framework models. These functions can be used when there are pre-trained models present and there is a need to use them along with other Vision functions. They can also be deployed in Real Time targets. The prerequisites for using the Deep Learning Inference Engines are:

  • Pre-trained models from supported libraries
  • The model must be a Frozen Model or a Saved Model
Note Model Training is currently not supported with Vision Development Module.