HSI Studio Documentation

HSI Studio opens hyperspectral datacubes, lets you explore and annotate them, and trains models on them. This is the manual.

If you've never worked with hyperspectral data before, start with What is hyperspectral imaging? — it explains how a hyperspectral image differs from an ordinary photo and what it's used for. Otherwise, jump to Getting Started.

Web or desktop

HSI Studio comes in two forms:

	Web	Desktop
Where	hsistudio.online	Windows / macOS / Linux
Cost	Free, no account	Licensed
Your data	Never leaves the browser	Never leaves the machine
Viewer, spectral plots, PCA, K-Means, Band Math	Yes	Yes
Annotation, ROI tools, export	Yes	Yes
Dataset Library and batch processing	—	Yes
ML Hub — training deep models	—	Yes
Kaggle sync and remote GPU training	—	Yes
Opens files from disk in place	—	Yes

Both do all their work locally. Your datacubes are never uploaded to an HSI Studio server — the web version runs entirely inside your browser, and the desktop version on your own machine. The only exception is a Kaggle training run, which you start deliberately and which sends the dataset to your own Kaggle account.

Pages marked desktop below cover features the browser version doesn't have.

The basics

Getting Started — installing, the file formats HSI Studio reads, and loading your first datacube
The Viewer — moving through bands, RGB and false-colour composites, colormaps, contrast, and every keyboard shortcut
Analysis — reading spectral signatures, PCA, K-Means clustering, Band Math and vegetation indices
Annotation & ROIs — the brush, the magic wand, classes, background removal, cropping and regions of interest
Exporting — every output format and what's inside it
Preferences — defaults and settings

Working at scale (desktop)

The Library — importing a folder as a dataset and running operations across all of it at once
ML Hub — training a model on your own data
Kaggle Integration — using free cloud GPUs when your machine doesn't have one

Getting help

The documentation inside the app — the same pages you are reading now, available offline while you work.

The documentation is available three ways, all identical:

Online at hsistudio.online/documentation
Inside the app — Help → Documentation, or press F1
On the landing page, under Looking for the manual?

Licence

HSI Studio is released under the PolyForm Noncommercial License 1.0.0. You may use it for personal, research, educational and nonprofit purposes. Commercial use is not permitted.

All pages

Getting StartedInstall HSI Studio or open it in your browser, then load your first hyperspectral datacube. The ViewerNavigate spectral bands, build RGB composites, apply colormaps, and control contrast in the HSI Studio viewer. AnalysisSpectral signatures, PCA, K-Means clustering, Band Math, and in-browser classification in HSI Studio. Annotation & ROIsLabel pixels with the brush, polygon, lasso and magic wand tools, manage classes, and extract regions of interest. The LibraryImport dataset folders, browse thumbnails, and run batch operations across hundreds of datacubes. ML HubTrain deep hyperspectral models — 3D-CNN, HybridSN, SpectralFormer, Mamba and custom architectures — on your own datasets. Kaggle IntegrationSync datasets to Kaggle and train hyperspectral models on free cloud GPUs from inside HSI Studio. ExportingEvery output format HSI Studio can write — project files, datacubes, masks, images and CSV. PreferencesSet viewer defaults, dataset scan behaviour, ROI export conventions, and UI scale.