Deep Learning for Multi Crop Classification using Hyperspectral Data Abhijeet Ghodgaonkar DA-IICT July 6, 2018
Abstract
To perform Multi Crop Classification using CNN Hyperspectral data has both spatial and spectral components making a datacube which is very large to process. We have to adopt dimensionality reduction techniques and fea- ture mapping techniques in order to train a model that can get accurate crop classifica- tion. A Hyperspectral sensor collects information which is a set of narrow wavelength range of the electromagnetic spectrum called as a spectral band. Since it is in the hy- perspectral range, there will be many bands collected by the sensors . Some bands can be eliminated and reduced without affecting the classification result. These images are combined to form a three-dimensional (x, y, λ) hyperspectral data cube for processing and analysis.