Skip to content
For Self-Managed AI Platform users running v11.0, see the on-premise platform documentation
DataRobot API resources > AI accelerators > Use cases and horizontal approaches > Label data with the data annotator app

Label data with the data annotator app

Access this AI accelerator on GitHub

High-quality training data is necessary for a top-notch machine learning model. But how you can quickly and easily collect labels from a team of human reviewers? One way is to stand up a Flask app for quick labeling review. This notebook will show you how to leverage the data-annotator app to both (1) label new data and (2) label predicted data within an active learning situation after training a model with DataRobot.

The data-annotator app requires two inputs:

  • img_path: The app is currently configured for labeling images (jpg and png are both supported formats). You need to place these images within a directory and specify that path to the app.
  • data_path: You need to tell the app all possible labels for your images.
  • If you are classifying images that have not yet been labeled, you can provide a csv file with at least one column named label that contains all potential classes. See Scenario 1 below for more details.
  • If you are classifying images that have already been assigned labels, you can provide a csv file with at least two columns named img_path (filename of the image) and label (assigned class for the image).
  • If you are classifying images that have already been scored within DataRobot, please refer to Scenario 2 below for more details on how to configure the dataset.