HYPSO-1 Sea-Land-Cloud Labeled Dataset

A diverse dataset of 179 calibrated Hyperspectral Images for Earth and ocean observation, including pixel-level labels for 38 images for sea/land/cloud categories.

The images were taken from the HYPSO-1 satellite between June and December 2022, with a few additional images from February and March 2023.

There are three types of files available: HYPSO-1 image files in level L1A format as .nc data type, sea/land/cloud label files for the images that have them as .bin data type and preview images in .png data format. The trained segmentation model used in the related publication is also added as an .h5 file.
--------
This dataset is a reformatted version of the dataset first published in 2023. The radiance data files are removed, and the raw data is stored in netCDF4/HDF5 files in a compressed format, decreasing the total size from about 130GB to about 17GB.

Users are encouraged to use the hypso python package to generate up to date radiance data from the L1A data files.