The sugar dataset - A multimodal hyperspectral dataset for classification and research

Melchert, F. (Creator), Matros, A. (Creator), Biehl, M. (Creator), Seiffert, U. (Creator), University of Groningen, 2016



The sugar dataset is a multimodal hyperspectral dataset of sugar and sugar related substances.
The substances that were used for the creation of dataset are:
- Sugar Ester S170
- Sugar Ester S770
- Sugar Ester S1570
- Sugar Ester P1570
- D-Mannitol
- D-Sorbitol
- D-Glucose
- D-Galactose
- D-Fructose
All of the substances were hyperspectrally recorded using different sensors, namely:
- Canon EOS 70D
- ASD FiledSpec 3
- Neo VNIR-1600
- Neo VNIR-1800
- Neo SWIR-320m-e
- Neo SWIR-384
- Nuance Ex
The different sensors cover different wavelength ranges as well as different wavelength resolutions. This creates a unique dataset, that not only takles the question of hyperspectral classification, but also enable the research on topics like high dimensional data exploration, sensor invariant classification and dimensionality reduction.
Date made available2016
PublisherUniversity of Groningen
Temporal coverage14-Mar-2016 - 22-Jun-2016
Access to the dataset Open

    Keywords on Datasets

  • Dimensionality Reduction, Hyperspectral Data, Model invariant classification, Model Transfer
Related Publications
  1. The sugar dataset: A multimodal hyperspectral dataset for classification and research

    Melchert, F., Matros, A., Biehl, M. & Seiffert, U., Jul-2016, Machine Learning Reports: MIWOCI Workshop 2016. Schleif, F-M. & Villmann, T. (eds.). Bielefeld: Univ. of Bielefeld, Vol. 03. p. 15 18 p.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademic

View all (1) »

ID: 48325141