Hyperspectral Remote Sensing Datasets: Indian Pines, Pavia University, Botswana and Salinas

This collection comprises four widely used benchmark hyperspectral remote sensing datasets: Indian Pines, Pavia University, Botswana, and Salinas. These datasets are frequently employed in the development and evaluation of hyperspectral image analysis methods, particularly for land cover classification, dimensionality reduction, and feature extraction.Indian Pines: Acquired by the AVIRIS sensor over agricultural fields in northwest Indiana, USA. The dataset consists of 224 spectral bands (0.4\u20132.5 \u00b5m) with 145 \u00d7 145 spatial pixels, primarily covering agricultural and forested land with 16 ground-truth classes.Pavia University: Collected using the ROSIS sensor over an urban area in Pavia, Italy. It contains 103 spectral bands with 610 \u00d7 340 pixels, representing diverse urban land-cover classes such as asphalt, meadows, trees, and buildings.Botswana: Acquired by the Hyperion sensor aboard NASA\u2019s EO-1 satellite over the Okavango Delta, Botswana. The dataset includes 242 bands with 1476 \u00d7 256 pixels, capturing a complex wetland ecosystem with 14 land-cover classes.Salinas: Captured by the AVIRIS sensor over Salinas Valley, California, USA. It consists of 224 spectral bands with 512 \u00d7 217 pixels, dominated by agricultural fields with 16 classes, including vegetables, bare soils, and vineyards.These datasets are widely recognized benchmarks in hyperspectral imaging research and serve as standard references for evaluating classification, band selection, and feature extraction algorithms.