Graphical user interface

The TwinCAT 3 Machine Learning Model Manager is the central UI for editing ONNX files. The tool is integrated in Visual Studio and can be opened via the menu bar under TwinCAT > Machine Learning > Machine Learning Model Manager.

Graphical user interface 1:

Required Visual Studio version

The graphic interface of the TwinCAT 3 Machine Learning Model Manager is compatible with Visual Studio 2017, 2019, 2022 and the TcXaeShell.

Creating the JSON and PlcOpenXml

1. Open the Convert Tool tab.
Click Select files to open the file browser.
2. Select ONNX files (multi-select possible by Ctrl-clicking).
Selected ONNX files are listed on the left-hand side with their path and file name.
3. If required, you can select files and remove them from the list again by clicking the Remove selected from list button.
4. Click on Convert files to create the required JSON and PlcOpenXml.
The files are stored in the converted file path. The default path is <TwinCATPath>\Functions\TF38xx-Machine-Learning\ConvertToolFiles.
5. Click on Open target path to open the converted file path in the file browser.
The path can be changed with Select target path. The change is retained even after restarting the PC.
Graphical user interface 2:

The tool described here is also the central tool for managing ONNX files for the TwinCAT Machine Learning Inference Engine and TwinCAT Neural Network Inference Engine. For this reason, description files for the optional execution of the AI model are also generated with these products if the ONNX is compatible. The Convert to *.xml drop-down menu is relevant for these products. This is not relevant for the TwinCAT Machine Learning Server.

A pop-up window indicates the TwinCAT product with which the ONNX is compatible for each converted ONNX.

Graphical user interface 3:

Creating Custom attributes

A JSON can be selected via Select File and then edited. After editing, the original file is overwritten using Save changes.

Graphical user interface 4:

The Custom attributes are edited using the buttons: