Skip to main content

Overview

Clarity brings hyperspectral imagery, spectral analysis, and machine-learning workflows into one cloud platform. Inspect a scene, compare spectral signatures, prepare versioned training data, and apply trained models to new imagery.

Choose your workflow​

Start withWork towardGuides
ImageryA useful visualization and an understood spectral signalImport files, create color composites, and inspect spectra
Labels or reference spectraA versioned dataset and an evaluated modelPrepare datasets, train models, and apply them to new imagery
A repeatable taskAn automated or agent-assisted workflowPython SDK, Clarity AI, and agent documentation

Use Projects to keep related files, workspaces, datasets, models, and spectral libraries in one working context.

From imagery to a model result​

Toulouse hyperspectral scene shown as a color composite on the left and a categorical classification output on the right.

The Toulouse example pairs a color composite with a classification output. For the evaluation protocol and interpretation of this qualitative example, see Toulouse dataset evaluation.

  1. Upload data, confirm its processing level, and open it in Spectral Explorer.
  2. Inspect the imagery and create finalized labels or reference spectra.
  3. Create and finalize a dataset version.
  4. Choose a model type, train it, and inspect held-out metrics and predictions.
  5. Run inference on compatible imagery and validate the result for the intended application.

For a guided example, follow the plastics unmixing tutorial.

Extend your workflow​

Release information and help​

What's new records release availability. Guides mark Preview and Alpha features where they affect a workflow; their availability depends on your environment.

For access and assistance, see Account access or contact support@metaspectral.com.