Hyperia
HyperIA

Datasets

Create and validate HyperIA datasets from per-class CSVs before training.

HyperIA datasets group all labels (classes) you need for training. Each class is backed by a CSV where each row is one pixel: reflectance or features across bands, consistent with your multispectral or hyperspectral cube.

Creating a dataset: give it a name, then add labels via + Add label. You can upload several CSVs in one go; label names follow the file names. Requirements shown in the app: .csv extension, header row, and matching columns across files.

Validation: open the dataset to see band count, sample counts, band range, and a spectral signature preview so you can catch issues before training.

HyperIA dataset detail: label list, dataset statistics, and spectral signature chart
Dataset overview: labels per class, sample totals, band range, and spectral signature preview.

For a full click-through, see Quick start.

Quick start

Sign in, create a dataset from CSV labels, run a training, and review metrics in HyperIA.

Trainings

Configure classifiers, selectors, and band ranges; then compare metrics across HyperIA runs.