Hyperspectral Dataset of Wine Grape Diseases in the Eastern Helan Mountain Foothills

The HVDHSD dataset, collected from the eastern foothills of the Helan Mountains in Ningxia, China, offers significant advantages for agricultural remote sensing and pathology research. It systematically integrates hyperspectral data covering multiple wine grape varieties and distinct plant organs (leaves, fruits, vines), establishing a unique foundation for cross-variety and cross-organ disease research. The rich multimodal data serves as an ideal platform for developing novel hyperspectral analysis methodologies, encompassing comprehensive pixel-level spectral data collected from entire leaves, multiple leaf combinations, and complex vine structures. This substantial data collection provides robust sample support for statistical analysis and deep model training, while the inclusion of large-scale field-collected imagery supports the development of real-time monitoring systems tailored for in-situ field scenarios. Regarding dataset characteristics, the current version employs a binary annotation scheme (Healthy/Diseased) to ensure label reliability across varying infection stages. Additionally, reflecting the stochastic nature of disease outbreaks under natural field conditions, there is an inherent distributional difference between healthy and diseased samples.