When using pre-annotations, annotators cannot see predictions

If annotators can’t see predictions or if you encounter unexpected behavior after you import pre-annotations into Label Studio, review this guidance to resolve the issues.

1. Ensure Correct Project Settings

In the Settings > Machine Learning section for your project, make sure that the following settings are configured:

• Enable Show predictions to annotators in the Label Stream and Quick View
• Select the relevant Model Version in the drop-down. If there is no drop-down menu visible, there might not be a model version listed for the pre-annotations, or there might be another issue happening.
• Disable the option to Reveal pre-annotations interactively, which requires manual action from annotators to display pre-annotated regions.

2. Verify ML Backend Connection

Ensure that your machine learning (ML) backend is running and properly connected to Label Studio. You can check the connection status in the project settings under the Machine Learning section. If the backend is not connected, predictions will not be generated or displayed.

3. Retrieve Predictions Manually

Sometimes, predictions might not be automatically retrieved. In the data manager for your project, select the tasks you want to get predictions for and use the "Retrieve predictions" option from the drop-down actions menu. This will send the selected tasks to your ML backend to fetch predictions.

4. Check Task Source for Predictions

Verify that the task source contains the predictions. You can do this by inspecting the task JSON to ensure it includes the predictions array with the necessary details. The JSON format should look something like this:

{
  "data": {
    "image": "/path/to/image.jpg"
  },
  "predictions": [
    {
      "model_version": "1.0",
      "result": [
        {
          "id": "result1",
          "type": "rectanglelabels",
          "from_name": "label",
          "to_name": "image",
          "value": {
            "x": 10,
            "y": 20,
            "width": 30,
            "height": 40,
            "rectanglelabels": ["Label1"]
          }
        }
      ]
    }
  ]
}

5. Ensure Correct Labeling Configuration

Make sure that the labeling configuration in your project matches the structure of the predictions. The from_name and to_name fields in the predictions must correspond to the names defined in your labeling configuration.