Skip to main content
Version: v1.2.0

Use cases

H2O Hydrogen Torch enables novice and expert data scientists to solve a large set of diverse problem types in computer vision, natural language, and audio. Below explore a few use case examples that you can solve using H2O Hydrogen Torch.

Image classification/regression​

As example use cases, you can build an image classification or regression model in H2O Hydrogen Torch to:

  • Identify pneumonia on chest X-rays
  • Classify the type of landscape from satellite or drone images
  • Predict the sum of coins in images

Image object detection​

As example use cases, you can build an image object detection model in H2O Hydrogen Torch to:

  • Find abnormalities on chest X-ray images of pneumonia patients
  • Detect vehicles from traffic or drone cameras
  • Detect required pieces during an assembly process

Image semantic segmentation​

As example use cases, you can build an image semantic segmentation model in H2O Hydrogen Torch to:

  • Cut out objects from image backgrounds
  • Segment a series of video frames based on a video captured by a dashcam
  • Locate the exact form of an object in medical images

Image instance segmentation​

As example use cases, you can build an image instance segmentation model in H2O Hydrogen Torch to:

  • Segment individual cars, pedestrians, or other objects from a dashcam video
  • Detect and delineate distinct objects of interest in biological images

Image metric learning​

As example use cases, you can build an image metric learning model in H2O Hydrogen Torch to:

  • Search for identical products on an e-commerce website
  • Search for images with similar landmarks in a database

Text classification/regression​

As example use cases, you can build a text classification or regression model in H2O Hydrogen Torch to:

  • Predict customer satisfaction from transcribed phone calls
  • Categorize incoming emails from support@your-company.com and forward them to the appropriate department

Text token classification​

As example use cases, you can build a text token classification model in H2O Hydrogen Torch to:

  • Extract entities such as drugs or diseases from medical text
  • Named entity extraction (Name, Location, etc.)

Text span prediction​

As example use cases, you can build a text span prediction model in H2O Hydrogen Torch to:

  • Find relevant information from medical transcripts
  • Build a company-specific question-answering system

Text sequence to sequence​

As example use cases, you can build a text sequence to sequence model in H2O Hydrogen Torch to:

  • Simplify text which contains domain-specific terms
  • Summarize text for better understanding of its content

Text metric learning​

As example use cases, you can build a text metric learning model in H2O Hydrogen Torch to:

  • Detect fake reviews which are similar to each other
  • Find similar questions in a user forum to remove duplicates

Audio classification/regression​

As example use cases, you can build an audio classification or regression model in H2O Hydrogen Torch to:

  • Detect bird and frog species by using tropical audio recordings
  • Predict the number of different species in audio files
Feedback