Skip to main content
Version: v0.3.0

Tutorial 3B: Annotation task: Object detection

Overview

This tutorial describes the process of annotating and specifying an annotation task rubric for an object detection annotation task. To highlight the process, we are going to annotate a dataset that contains images of cars and coffee. This tutorial also quickly explores how you can download the fully annotated dataset supported in H2O Hydrogen Torch.

Step 1: Explore dataset

We are going to use the preloaded car-or-coffee-demo demo dataset for this tutorial. The dataset contains 40 images, each depicting a car or coffee. Let's quickly explore the dataset.

  1. On the H2O Label Genie navigation menu, click Datasets.
  2. In the datasets table, click car-or-coffee-demo.

Multiple objects: Leftmost side is a parked red volkswagen vehicle, to the right is a red cup of coffee, next object is a mug raised with a sunrise as background, the rightmost object is Formula 1 race car

Step 2: Create an annotation task

Now that we have seen the dataset let's create an annotation task that enables you to annotate the dataset. An annotation task refers to the process of labeling data. For this tutorial, an object detection annotation task refers to specifying one or more object classes (labels) to each input image. Let's create an annotation task.

  1. Click New annotation task.
  2. In the Task name box, enter tutorial-3b.
  3. In the Task description box, enter Annotate a dataset containing images of cars and coffee.
  4. In the Select task list, select Object detection.
  5. Click Create task.

Step 3: Specify annotation task rubric

Before we can start annotating our dataset, we need to specify an annotation task rubric. An annotation task rubric refers to the labels (for example, object classes) you want to use when annotating your dataset. For our dataset, there are two labels (object classes) we want to specify, car and coffee. Let's define the annotation task rubric.

  1. In the New object name box, enter Car.
  2. Click Add.
  3. Click Add object class.
  4. In the New object name box, enter Coffee.
  5. Click Add.
  6. For the Coffee object color, choose an option (different from the one preselected for the Car object).
  7. Click Continue to annotate.

Annotation task rubric configuration of Car and Coffee classes

Step 4: Annotate dataset

Now that we have specified the annotation task rubric, let's annotate the dataset (the first image).

  1. A zero-shot learning model is on by default when you annotate an object detection annotation task. The model accelerates the annotation (labeling) process by capturing the object observed in the image with a bounding box. In particular, the zero-shot model selects the fitting label while indicating where you can observe the label object in the image. In this case, "fitting" refers to the zero-shot learning model's calculations.

    You can immediately start annotating in the Annotate tab or wait until the zero-shot model is ready to provide annotation suggestions. H2O Label Genie notifies you to Refresh the instance when zero-shot predictions (suggestions) are available.

    Refresh

    For example, in this tutorial, the model selected the Car label while indicating where the car is located in the image.

    Annotated dataset with selection of Car class

    info
    • To learn about the utilized model for an object detection annotation task, see Zero-shot learning models: Object detection.
    • During the annotation process of an image object detection dataset, you can download generated zero-shot predictions in the Export tab. To download all generated zero-shot predictions, consider the following instructions:
      caution
      • If the Enable zero-shot predictions setting is turned Off, the zero-shot learning model utilized for an object detection annotation task is not available during the annotation process while preventing the generation of zero-shot predictions. To turn On the Enable zero-shot predictions setting, see Enable zero-shot predictions.
      • The time it takes H2O Label Genie to generate zero-shot predictions depends on the computational resources of the instance.
      1. Click the Export tab.
      2. In the Export zero-shot predictions list, select Download ZIP.

    If the zero-shot learning model did not accurately capture the object on the image, inside the bounding box, double-click inside the bounding box. Double-clicking inside the bounding box enables you to move and resize the bounding box to accurately cover the area the car (object) takes on the image.

  2. Click Save and next.

    Note
    • Save and next saves the annotated image
    • To skip an image to annotate later: Click Skip.
      • Skipped images (samples) reappear after all non-skipped images are annotate
    • To download all annotated samples so far, consider the following instructions:
      1. Click the Export tab.
      2. In the Export approved samples list, select Download ZIP.
        note

        H2O Label Genie downloads a zip file containing the annotated dataset in a format that is supported in H2O Hydrogen Torch. To learn more, see Downloaded dataset formats: Object detection.

Download annotated dataset

After annotating all the images, you can download the dataset in a format that H2O Hydrogen Torch supports. Let's download the annotated dataset.

  1. In the Annotate tab, click Export approved samples. Notification of completed annotation task ready to export approved samples
  2. In the Export approved samples list, select Download ZIP.

Summary

In this tutorial, we learned the process of annotating and specifying an annotation task rubric for an object detection annotation task. We also learned how to download a fully annotated dataset supported in H2O Hydrogen Torch.

Next

To learn the process of annotating and specifying an annotation task rubric for other various annotation tasks in computer vision (CV), natural language processing (NLP), and audio, see Tutorials.


Feedback