Hyperspectral Open Set Classification towards Deep Networks Based on Boxplot

Abstract Recently, hyperspectral imaging (HSI) supervised classification has achieved an astonishing performance by using deep learning. However, most of them take the ideal assumption ‘closed set’, where all testing classes have been known during training. In fact, in real world, new unseen training may appear testing. Obviously, traditional methods cannot operate correctly which requires classifiers not only to classify classes, but reject unknown order avoid false positives. This challenge is called ‘open set classification’(OSC). Considering increased applications learning rejecting vital importance. To tackle it, we present a simple effective HSI OSC method toward networks. this method, tighten decision boundaries SoftMax function last layer networks boxplots analysis statistical characteristics probability distribution and generate proper rejection threshold for each class. test proposed experiments are conducted on three datasets. The results show that outperforms existing state-of-the-art methods.