Skip to main content

First Image & Recipe Test (Repeat to creating first recipe)

This guide helps you capture your first image, configure basic Image Settings, and test a simple classifier recipe using the OV20i, following the official user manual steps.

Prerequisites​

  • Camera powered and connected (all 4 LEDs are green)
  • You can access the camera's UI (e.g., http://192.168.0.100)
  • Sample part placed under the camera's view
  • Proper lighting is configured

Step 1: Create & Activate a New Classifier Recipe​

  1. In the UI, go to Recipes > All Recipes.
  2. Click + New Recipe, name it (e.g., "TestRecipe"), and select Classifier type.
  3. On the new recipe listed, click Actions > Activate, then hit Activate.

Step 2: Configure Image Settings​

  1. Click Edit then Open Editor.
  2. Select Configure Imaging (bottom left of editor).
  3. Adjust:
    • Focus (slider or manual value)
    • Exposure
    • LED Light Pattern (top/down, brightness)
    • Gamma for contrast
  4. Click Save Imaging Settings to apply.

Step 3: Template Image & Alignment​

  1. Navigate to Template Image and Alignment.
  2. Capture the current view as the template.
  3. For now, click Skip Aligner (optional).
  4. Click Save.

Step 4: Inspection Setup – Create ROI​

  1. Navigate to Inspection Setup.
  2. Add your part or defect class.
  3. Draw the Region of Interest (ROI) around the relevant area.
  4. Click Save to apply.

Step 5: Train the Classifier​

  1. Go to Classification block in the Recipe Editor.
  2. For each class (e.g., Good/Bad), capture ≥5 diverse images.
  3. Click Train Classification Model.
  4. In the prompt, choose Fast (trial) or Accurate (production).
  5. Adjust iterations as needed.
  6. Click Start Training.
  7. When done, click Live Preview to test the model.

Step 6: Evaluate Results​

  • "PASS" appears when a correct part is shown.
  • "FAIL" appears for other objects or defective parts.
  • Use Live Preview to watch real-time inference.

Troubleshooting Tips​

  • No video feed? → Check Ethernet and LEDs.
  • Capture disabled? → Ensure Imaging settings are saved.
  • ROI not applying? → Redraw and click Save.
  • Model not training? → Verify all classes have ≥5 labeled samples.

🔗 See Also​