Please ensure Javascript is enabled for purposes of website accessibility
Powered by Zoomin Software. For more details please contactZoomin

AVEVA™ Vision AI Assistant

Retrain a skill

  • Last UpdatedFeb 12, 2024
  • 2 minute read

After providing feedback for individual images, you can then choose to retrain the skill. After retraining, you can compare the current skill to the retrained skill based on the metrics and choose the better performing skill.

To retrain a skill

  1. On the Review page, select Retrain.

    The Train Your AI page appears. The card displays the count of trained and untrained images.

    The following is for an Anomaly Detection skill:

    Embedded Image (65% Scaling) (LIVE)

    The following is for a Discrete State Detection skill:

    Embedded Image (65% Scaling) (LIVE)

  2. Optional: Import additional image datasets.

  3. Select Train AI.

    The Review page appears and the status of the skill changes to Retrained.

  4. From the Review page, you can compare the metrics between the current skill and retrained skill.

    The following is for an Anomaly Detection skill when the current skill is selected:

    The following is for an Anomaly Detection skill when the retrained skill is selected:

    The following is for a Discrete State Detection skill when the current skill is selected:

    The following is for a Discrete State Detection skill when the retrained skill is selected:

  5. Select Review to provide feedback for the image classifications again.

  6. To deploy the retrained version of the skill, select the retrained version, and then select Deploy.

    The retrained version replaces the current (deployed) version of the skill, and details pertaining to the current version are lost.

  7. To keep the current version and ignore the retrained version of the skill:

    1. Select Reset.

    2. On the Reset Retrain Data dialog, select Confirm.

TitleResults for “How to create a CRG?”Also Available in