A research review on deep learning combined with hyperspectral Imaging in multiscale agricultural sensing

Efficient and automated data acquisition techniques, as well as intelligent and accurate data processing and analysis techniques, are essential for the advancement of precision agriculture. Hyperspectral images have the capability to capture both spatial and spectral features of an object's surface. Deep learning, as a powerful technique for extracting features from hyperspectral data, has shown promising results in multi-scale agricultural sensing and management. Despite the significant progress made in deep learning research, there are still many unresolved questions and aspects that require further exploration. This review aims to provide an overview of the application of deep learning combined with hyperspectral imaging in multiscale agricultural management. It focuses on the general aspects of deep learning techniques for processing multiscale hyperspectral agricultural data, including commonly used models, the main challenges that need to be addressed, and the existing research gaps. Furthermore, potential solutions and future research directions are proposed to enhance the relevance of these techniques in real-world applications. It should be noted that this review solely concentrates on food and crop scopes, excluding animal-related research literature at present.