Skip to main content
Version: v0.4.0

Supported annotation tasks

Overview​

H2O Label Genie supports various annotation tasks in computer vision (CV) (image), natural language processing (NLP) (text), and audio.

note

To learn how to create an annotation task, see Create an annotation task.

Text annotation tasks​

Text classification​

  • Description: A text classification annotation task specifies one or more categorical target labels for each input text.
note
  • H2O Label Genie supports multi-label text classification annotation tasks.
  • To accelerate the labeling process for text classification annotation tasks, H2O Label Genie offers the ability to utilize a zero-shot learning model. To learn more, see Zero-shot learning models.

Text regression​

  • Description: A text regression annotation task specifies one continuous target label for each input text.

Text-entity recognition​

  • Description: A text-entity recognition annotation task specifies one or more defined entities for each unstructured input text.

Text summarization​

  • Description: A text summarization annotation task specifies a summary for each input text.
note

To accelerate the labeling process, H2O Label Genie offers the ability to utilize pretrained text summarization models. To learn more, see Zero-shot learning models.

Text-generative AI​

  • Description: A text-generative AI annotation task specifies a particular text response for the text column(s).
note
  • Zero-shot learning models: To accelerate the labeling process for a text-generative AI annotation tasks, H2O Label Genie offers the ability to utilize several zero-shot learning models. To learn more, see Zero-shot learning models: Text-generative AI
    • The available zero-shot learning models for a text-generative AI annotation task are from the large language model (LLM) h2oGPT family
  • Bring your own model: For text-generative AI annotation tasks, H2O Label Genie enables you to bring your own model
    • To learn how to specify the path to your own model in H2O Label Genie, see Custom model URL

Image annotation tasks​

Image classification​

  • Description: An image classification annotation task specifies one or more categorical target labels for each input image.
note
  • H2O Label Genie supports multi-label image classification annotation tasks.
  • To accelerate the labeling process for image classification annotation tasks, H2O Label Genie offers the ability to utilize a zero-shot learning model. To learn more, see Zero-shot learning models.

Image regression​

  • Description: An image regression annotation task specifies one continuous target label for each input image.

Object detection​

  • Description: An object detection annotation task specifies one or more object classes (labels) for each input image.
note

To accelerate the labeling process for object detection annotation tasks, H2O Label Genie offers the ability to utilize a zero-shot learning model. To learn more, see Zero-shot learning models.

Image instance segmentation​

In H2O Label Genie, for an image instance segmentation annotation task, you need to assign one or more object classes (labels) to each input image.

note

To accelerate the labeling process for image instance segmentation annotation tasks, H2O Label Genie offers the ability to utilize a zero-shot learning model. To learn more, see Zero-shot learning models.

Audio annotation tasks​

Audio classification​

  • Description: An audio classification annotation task specifies one or more categorical target labels for each input audio.
note

H2O Label Genie supports multi-label audio classification annotation tasks.

Audio regression​

  • Description: An audio regression annotation task specifies one continuous target label for each input audio.
Feedback