KIT | KIT-Bibliothek | Impressum | Datenschutz

Spatially resolved ingredient detection in spice mixes using 3D convolutional neural networks

Anastasiadis, Johannes; Krippner, Wolfgang; Puente Léon, Fernando

Abstract (englisch):
Abstract A method using spectral information to detect substances in mixtures is given. The presented convolutional neural network is using three-dimensional convolutions to process hyperspectral images. Reflectance values can be fed directly into the network and are not preprocessed. Due to the architecture, the neural network performs a spatially invariant operation. Detection performance is demonstrated by a dataset containing spice mixtures.

Open Access Logo

Verlagsausgabe §
DOI: 10.5445/IR/1000092319
Veröffentlicht am 18.03.2019
Zugehörige Institution(en) am KIT Institut für Industrielle Informationstechnik (IIIT)
Publikationstyp Proceedingsbeitrag
Jahr 2019
Sprache Englisch
Identifikator ISBN: 978-3-7315-0864-9
KITopen-ID: 1000092319
Erschienen in OCM 2019 - Optical Characterization of Materials : Conference Proceedings. Ed.: J. Beyerer; F. Puente León, T. Längle
Verlag KIT Scientific Publishing, Karlsruhe
Seiten 35-44
Relationen in KITopen
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page