Interactive Deep Learning for Hyperspectral Images

Deep models have achieved remarkable results in many real-world applications based on quality-labeled training samples. However, accruing reliable training samples is expensive in many real-world applications, particularly in the remote sensing domain. Therefore, there is a need to develop interactive deep schemes that can help to attain reliable, informative, and heterogeneous samples for learning. For this Special Issue of Remote Sensing, we aim to present a collection of articles related to “Interactive Deep Learning for Hyperspectral Images” including, but not limited to, the following topics:

1. Investigation of the behavior and performance of deep models in terms of the computational cost and generalized performance for both on-shelf and novel classification methods under different experimental conditions.

2. Development of several novel strategies to limit the Hughes phenomenon for hyperspectral image classification by exploiting several on-shelf and novel sample selection methods. This includes the generation of a good spectral library, the development of suitable models to upscale the ground spectra to air and space-borne measurements, evolving application methodologies, and device algorithms for pre and post-processing hyperspectral cubes.

Dr. Muhammad Ahmad
Dr. Adil Mehmood Khan
Dr. Diego Oliva
Dr. Omar Nibouche
Guest Editors

More information: https://www.mdpi.com/journal/remotesensing/special_issues/hyperspectral_deep_models