Process

Based on the train results, process the image set to test the model and create predictions.

To test the trained tool:

  • Adjust the runtime parameters as needed. The parameters vary per tool.

    Note: For more information on tool-specific processing and runtime parameters, see Locator, Segmenter, or Classifier.

  • Click Process All.

After processing, markers appear on each view according to their assigned role, and assigned and predicted label. For more tool-specific information on processing and verifying, please see the individual tool topics.

Result Panel

After you complete the Process step, the predicted results for the complete input view set display as statistics and metrics.

Use the results to:

  • Identify images that the tool gave incorrect predictions to.

  • Identify outliers or the limitation of the current trained tool.

For more information on how to use the results, see Metrics and Statistics.

Completing Tool Configuration

After evaluating the results, you can:

  • Modify image labels.

  • Add more images to the training data set in the Label step.

  • Click Accept Label to accept the predictions.

Configuring a tool is an iterative process that requires repeating some or all four configuration steps multiple times. After reviewing the metrics and statistics, you may need to add more images, adjust the trained view set, adjust the train and runtime parameters. This is true in particular when working with larger view sets. Manually labeling Labeling is the process of marking features or defects in images, or categorizing the images into classes. Labeling is crucial for training Deep Learning Lite and Deep Learning Standard tools because the labels serve as training material to guide the tools how to function correctly. and assigning roles to each view is tedious and time consuming. Using the bootstrap labeling Bootstrap labeling is a fast labeling method, which involves labeling a small sample of images, training the tool, and reviewing the results. Accept views with correct markings to convert them into labels, and replace incorrect markings with correct labels. and auto-role assignment features make this process more efficient.