The conference https://rosus.feri.um.si/rosus2019/ will be held on March 21, 2019 in the premises of UM-FERI, Koroška cesta 46, Maribor auditorium G2-Beta
Talk by Dr Matthias Kerschhaggl, CTO of EVK
Author: Dr. Matthias Kerschhaggl, EVK
Title: State of the art hyperspectral data analysis and prospects with respect to AI
Multidimensional data such as stemming from hyperspectral imagery (HSI) with its high degree of collinearity need sophisticated multivariate methods to extract relevant features in various machine vision applications. Most of the current methods emphasize the spectroscopic nature of the data and focus less on other features such as object shape and texture also present in HSI-cubes. Corresponding features are often treated separately and sequentially rather than analysing the data following a more holistic approach. This limitation can be overcome by introducing e.g. deep learning methods (such as stacked autoencoders) to these combined spatio-spectral datasets.
This talk will give a brief introduction into state of the art hyperspectral analysis algorithms as well as an outlook into the benefits of modern AI methods related to this high potential machine vision technology