Pixel annotation in hyperspectral and very high resolution
(VHR) images has relied on supervised classifiers so far [1]–
[4]. However, manual selection and annotation of a sufficiently
large number of training pixels is unfeasible for very large
datasets. In order to handle large datasets, active learning
techniques have been proposed for selecting a small training
set that represents well not only all classes under annotation
but also discriminative samples on the boundary between
classes [1], [2], [5]–[8].