Classify Parts

Set up and use machine learning (ML) models to automatically classify parts under user-defined labels and into part sets.

The Classify tool uses the benefits of machine learning (ML) to auto-assign labels to parts based on their shape. Create and assign each label to parts so that the ML model learns what the parts look like. Once you have assigned labels, train a model and use it to classify parts.

Assign User-Defined Labels: To begin the process, assign user-defined labels. For example, if you want to divide a selection of vehicle parts into two groups, such as “panels” and “extrusions”, create those labels, and assign them to the parts with representative geometry. You can assign the “panels” label to the floor, roof, and door panels, and the “extrusions” label to the crash longitudinal.
Train a Model: After assigning the labels, train a classification model. Manually assigning a large number of parts to each label before training helps the model predict future labels.
Classify Parts: Classify parts that the ML model has never seen before. Once classification is complete, part sets are created that contain the predictions of the ML model. Decide how to use the resultant part sets, such as applying different meshing approaches for “panels” and “extrusions”.

Note: Since classification functionality use cases are diverse, example pre-trained ML models are not provided. Use the tools provided to create your own user-defined models.

From the Assembly ribbon, click the Classify tool.

Figure 1.

A secondary ribbon is displayed, which contains the following part classification tools:

Label

Create or update a classifier. A classifier contains user-provided label information to train machine learning models.

Train

Select an existing classifier to train a machine learning model.

Classify

Use an existing trained machine learning model to automatically classify parts under user-defined labels and into part sets.