Artificial Intelligence (AI) is set to play a key role in reducing the programming and engineering effort required to create automation solutions. It is also making control logic more agile and production processes more flexible and precise – especially in industry. For example, machine-learning algorithms help systems that perform visual quality checks in production plants or image-guided robot systems. These algorithms allow automation systems to react much more flexibly to unexpected situations and quality defects. The response to unexpected situations proceeds automatically during runtime. The professional knowledge – for example, regarding the color, consistency, or quality of a product or process – can be transmitted to automation. As a result, these systems operate much more efficiently.
AI Model Deployer
The AI Model Deployer is a Windows application that provides a simple interface for deploying trained Machine Learning models to Siemens targeted hardware. It provides many functionalities, such as:
Create projects based on application templates (Image Classification/Object detection for TM NPU), and keep track of the projects
Import Image Classification/Object detection models for the application
Convert and optimize trained neural networks for target inference hardware
Set up the application configuration, e.g. setting up the hardware, network and camera configurations
Build the application package for the target inference device (SD card image for TM NPU)
AI Model Deployer provides two ways to perform the task:
Validity of the documentation
This documentation is valid for the AI Model Deployer application to be used for the Industrial AI Visual Inspection solution. This version allows you to:
deploy image classification or object detection models to the SIMATIC S7-1500 TM NPU (6ES7556-1AA00-0AB0)
Further information can be found in the getting started examples on the Internet at the following link.
Convention
In this manual, the terms "host PC/host machine" refers to the PC or Virtual machine which has AI Model Deployer installed. The term "Application" refers to automation application in plant.
Download examples
In the following you will find the downloadable content referenced in the AI Portfolio Getting Started Portal. The content is tailored to helping an Automation engineer in setting up an initial AI application with Software for Artificial Intelligence components and consists of example code and an example project, as well as a prepared training environment in the form of a Jupyter notebook.
History
The following editions of these operating instructions have been published:
Edition |
Version |
Comment |
|---|---|---|
04/2022 |
V1.0 |
First edition |
09/2023 |
V1.1 |
New functions TM NPU firmware 2.0 adds FTP and debug function. AI Model Deployer supports TM NPU firmware 2.0. Projects created for TM NPU firmware 2.0 in AI Model Deployer support the following new features:
Removed function
|