Use case examples¶
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 through H2O Hydrogen Torch.
Note
- Future releases of H2O Hydrogen Torch will introduce more problem types.
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 an 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 video captured using a dashcam
-
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¶
For 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
- Submit and view feedback for this page
- Send feedback about H2O Hydrogen Torch to cloud-feedback@h2o.ai