Your AI can inspect the entire image, but this is often not the most accurate way of doing inspection. Generally, it is better to inspect sub regions of the image.

<aside> 💡 2 main concepts: Inspection Type and Inspection Regions. Each Inspection Type can have 100s of associated Inspection Regions. In the Classifier Block each Inspection Type can be given it’s own characteristics.


☑️ How to:

  1. Click on the Edit option under the ROI Block


  2. [Optional] Select the Add Inspection Type option to define the application(s). You can add as many different Inspection Types as you need for each use case and can custom-label the Inspection Type to match your company’s verbiage.

    1. Creating a new Inspection Type will automatically add a default Good and Bad class.



  3. Click on +Square Inspection Region to generate a square that can be moved and resized to accurately capture your selected Inspection Region.


  4. Modify each square around each ROI. You can copy and paste the square for the selected ROIs that are the same. In this example, there are nine small holes on the bracket. After you have created the 1st Inspection Region for Small 1, you can select the copy function, paste the square on Small 2, and the system will auto-generate the label Small 2. Repeat this process until all Inspection Regions have been marked

  5. Once you have identified all of the ROIs for one Inspection Type, you can repeat the same steps and accurately attribute the ROI to the correct Inspection Type. Click on the pull down menu from the Inspection Type column and choose the correct Inspection Type for the new Inspection Region.


  6. Save all of the settings once all of the ROIs have been accurately drawn, labeled and identified as the correct Inspection Type.

  7. Select the Live Preview Mode to confirm that all of the ROIs are being recognized by the device.

💡Tips & Best Practices

Here is an example of how to sub divide the image for maximum accuracy.

Untitled (10).png