61°50' N

A dataset of hyperspectral images of (eucalyptus) wood boards, for wood recognition (heartwood and sapwood). The following archives include the raw images of boards with corresponding annotated files (20200316_0.rar and 20200316_1.rar), and the extracted dataset (data_132_66.tar.gz), consisting of extracted cuboids of dimension 32x32x320, divided in training and test set respectively.

The dimension of the archives is of Given the dimensions of the files, the extracted dataset is available to download directl, Raw datasets 20200316_0.rar and 20200316_1.rar have dimension of 12GB and 16 GB respectively.

The archives can be downloaded from the following URLs:

Raw data 20200316_0.rar (12 GB): http://fesr1111-h2i.inf.unibz.it.s3-website-eu-west-1.amazonaws.com/20200316_0.rar Raw data 20200316_1.rar (16 GB): http://fesr1111-h2i.inf.unibz.it.s3-website-eu-west-1.amazonaws.com/20200316_1.rar Processed data (70 MB): http://fesr1111-h2i.inf.unibz.it.s3-website-eu-west-1.amazonaws.com/data_132_66.tar.gz

The raw data was provided by Microtec Gmbh to the Free University of Bozen-Bolzano, Faculty of Computer Science, in the context of the FESR project Id: 1111, H2I: Hyper-Spectral Images for Inspection Applications, Funding Scheme: EFRE-FESR 2014-2020.

The hyperspectral instrument DESIS (DLR Earth Sensing Imaging Spectrometer) is one of four possible payloads of MUSES (Multi-User System for Earth Sensing), which is mounted on the International Space Station (ISS). DLR developed and delivered a Visual/Near-Infrared Imaging Spectrometer to Teledyne Brown Engineering, which was responsible for integrating the instrument. Teledyne Brown designed and constructed, integrated and tested the platform before delivered to NASA. Teledyne Brown collaborates with DLR in several areas, including basic and applied research for use of data. DESIS is operated in the wavelength range from visible through the near infrared and enables precise data acquisition from Earth's surface for applications including fire-detection, change detection, maritime domain awareness, and atmospheric research. Three product types can be ordered, which are Level 1B (systematic and radiometric corrected), Level 1C (geometrically corrected) and Level 2A (atmospherically corrected). The spatial resolution is about 30m on ground. DESIS is sensitive between 400nm and 1000nm with a spectral resolution of about 3.3nm. DESIS data are delivered in tiles of about 30x30km. For more information concerning DESIS the reader is referred to https://www.dlr.de/eoc/en/desktopdefault.aspx/tabid-13614/

Description

The Global Hyperspectral Imaging Spectral-library of Agricultural crops (GHISA) is a comprehensive compilation, collation, harmonization, and standardization of hyperspectral signatures of agricultural crops of the world. This hyperspectral library of agricultural crops is developed for all major world crops and was collected by United States Geological Survey (USGS) and partnering volunteer agencies from around the world. Crops include wheat, rice, barley, corn, soybeans, cotton, sugarcane, potatoes, chickpeas, lentils, and pigeon peas, which together occupy about 65% of all global cropland areas. The GHISA spectral libraries were collected and collated using spaceborne, airborne (e.g., aircraft and drones), and ground based hyperspectral imaging spectroscopy.

The GHISA for the Conterminous United States (GHISACONUS) Version 1 product provides dominant crop data in different growth stages for various agroecological zones (AEZs) of the United States. The GHISA hyperspectral library of the five major agricultural crops (e.g., winter wheat, rice, corn, soybeans, and cotton) for CONUS was developed using Earth Observing-1 (EO-1) Hyperion hyperspectral data acquired from 2008 through 2015 from different AEZs of CONUS using the United States Department of Agriculture (USDA) Cropland Data Layer (CDL) as reference data.

GHISACONUS is comprised of seven AEZs throughout the United States covering the major agricultural crops in six different growth stages: emergence/very early vegetative (Emerge VEarly), early and mid vegetative (Early Mid), late vegetative (Late), critical, maturing/senescence (Mature Senesc), and harvest. The crop growth stage data were derived using crop calendars generated by the Center for Sustainability and the Global Environment (SAGE), University of Wisconsin-Madison.

Provided in the CSV file is the spectral library including image information, geographic coordinates, corresponding agroecological zone, crop type labels, and crop growth stage labels for the United States.

In support of U.S. Geological Survey (USGS) Southwest Biological Science Center researchers, and in coordination with the Bureau of Land Management (BLM) and National Ecological Observatory Network (NEON), the USGS National Uncrewed Systems Office (NUSO) conducted uncrewed aircraft systems (UAS) remote sensing flights over two BLM Assessment, Inventory, and Monitoring (AIM) plots at the NEON Moab site in Utah for multi-scale carbon sequestration research on public lands. The UAS data collected include natural color, multispectral, and hyperspectral imagery, and lidar to capture diverse information about vegetation and soils on drylands. The first site (“site 1”) features intact sagebrush and was mapped on May 3, 2023. The second site (“site 7”) is located on a grazed rangeland environment and was mapped on May 5, 2023. These UAS surveys were conducted in early May 2023 to coincide spatially and temporally with ground-based BLM AIM sampling and airplane-based remote sensing surveys by NEON. This portion of the data release presents hyperspectral data products from low-altitude UAS flights at two dryland sites approximately 40 km south of Moab, Utah. A Headwall Nano-Hyperspec line scanning sensor was flown at an altitude of 31 meters above ground level on a DJI Matrice 600 Pro UAS with approved government edition firmware. The hyperspectral images were post-processed using the sensor manufacturer's proprietary software and following their recommended workflow. The orthorectified hyperspectral reflectance images are stored as 32-bit single precision floating point numbers in flat binary files with a band sequential (BSQ) interleave. Each image is accompanied by an ASCII text header file (.hdr) containing band center wavelengths and other parameters relevant to the images. Each image has 274 spectral bands spanning the visible and near infrared wavelengths, 398 to 1002 nm. The images were georeferenced to a geographic coordinate system (latitude and longitude) and WGS84 datum with spatial resolution 1.7 cm (site 1) and 1.5 cm (site 7). There are 30 hyperspectral images with accompanying header files captured at site 1, provided in 5 zip folders to facilitate bulk download. There are 33 hyperspectral images with accompanying header files captured at site 7, provided in 5 zip folders.

The repository contains four PRISMA sensor hyperspectral simulated dataset, corresponding to two different scenes, P1 and P4. For each scene there are 2 different dataset:one with simulated random noisesone with simulated random noises ans simulated striping noise.Each dataset contains 256 spectral bands.Each dataset comes with its ENVI compliant headerData are in radiance, and data units are W/(sr*m2 *nm)More info can be found on the readme.txt file.

Recent advancements in hyperspectral imaging (HSI) for early disease detection have shown promising results, yet there is a lack of validated high-resolution (spatial and spectral) HSI data representing the responses of plants at different stages of leaf disease progression. To address these gaps, we used bacterial leaf spot (Xanthomonas perforans) of tomato as a model system. Hyperspectral images of tomato leaves, validated against in planta pathogen populations for seven consecutive days, were analyzed to reveal differences between infected and healthy leaves. Machine learning models were trained using leaf-level full spectra data, leaf-level Vegetation index (VI) data, and pixel-level full spectra data at four disease progression stages. The results suggest that HSI can detect disease on tomato leaves at pre-symptomatic stages and differentiate bacterial disease spots from abiotic leaf spots.

The data available consist of airborne hyperspectral imagery acquired for the Mer Bleue Arctic Surrogate Simulation Site (MBASSS) S2/L8 Data Product Validation Project in 2016. MBASSS was a collaborative effort aimed at developing a systematic approach for ongoing assessment and validation of satellite based land information products from Landsat 8 OLI and Sentinel 2 satellites. The airborne systems used for this project were the CASI-1500 and SASI-644 hyperspectral instruments (ITRES Research, Calgary AB) installed in the National Research Council Canada Flight Research Lab (NRC-FRL) Twin Otter aircraft. Standard level 2 processed imagery is provided for download as rasters in ENVI Standard format. Imagery is available from April 20, May 11, May 24 and June 23, 2016 as a set of 12 individual flight lines per date. The imagery has been atmospherically corrected and during the geocorrection process, it has been resampled to 1 m pixel size. Currently CASI and SASI imagery are provided separately. Metadata for each flight line is provided in external ascii ENVI header files (*.hdr) and *.met files. The geocorrected imagery provided with pixel level information including pixel view zenith angle (off Nadir angle), DEM, view azimuth angle, radiance path distance, column and row numbers of pixels in non-geocorrected image file, and relative pixel offset between calculated and assigned pixel location. This information is provided in associated *.nad and *.nad.hdr files.

Airborne remote-sensed hyperspectral in-situ radiometry data and hyperspectral imagery collected by the NERC Field Spectroscopy Facility (FSF) Headwall Co-aligned VNIR and SWIR imager (450-2500 nm) with LiDAR instruments mounted on a drone platform. These hyperspectral data collected over a sandy and rocky shore have associated uncertainty estimations that will be used to develop of radiometric proxies for plastics detection and assess future mission requirements. This dataset was collected on 29th September 2020 at Tyninghame beach, East Lothian, Scotland using a range of different plastic targets.

Corescan© Hyperspectral Core Imager Mark III (HCI-III) system data were acquired for hand samples, and subsequent billets made from the hand samples, collected during the U.S. Geological Survey (USGS) 2014, 2015, and 2016 field seasons in the Nabesna area of the eastern Alaska Range. The HCI-III system consists of three different components. The first is an imaging spectrometer which collects reflectance data with a spatial resolution of approximately 500 nanometers (nm) for 514 spectral channels covering the 450-2,500 nm wavelength range of the electromagnetic spectrum (Martini and others, 2017). The second is a spectrally calibrated RGB camera that collects high resolution imagery of the samples with a 50 micrometer (μm) pixel size. The third component is a three-dimensional (3D) laser profiler that measures sample texture, surface features and shape with a vertical resolution of 20 μm (Martini and others, 2017). Corescan reflectance data were provided for a total of 63 hand samples and four billets analyzed using the HCI-III system in three scans.

The FSSCat collection provides hyperspectral data coverage over a number of locations around the world, as measured by the HyperScout 2 sensor. The FSSCat hyperspectral data products are comprised of 50 spectral bands, covering a spectral range of 450 – 950 nm with a spectral resolution of 18 nm (at FWHM). Imagery is available with an along-track ground sampling distance (GSD) of 75 m. To ensure a high degree of radiometric accuracy, HyperScout 2 data are validated through comparison with Sentinel-2 data products. The processing level of the data is L1C – calibrated top-of-atmosphere radiance, reflectance or brightness temperature. The raster type of the L1C data product is a GRID – a 2D or 3D raster where the (geo)location of the data is uniquely defined by the upper left pixel location of the raster and the pixel size of the raster, and the projection parameters of the raster (if georeferenced). The third dimension can e.g. be a spectral or third spatial dimension. The L-1C VNIR data product includes a hyperspectral cube of TOA reflectance in the VNIR range, as well as relevant meta-data that adheres to EDAP's best practice guidelines. This product consists of georeferenced and ortho-rectified image tiles that contain spectral reflectance data at the top-of-the-atmosphere. Each image tile contains radiometrically corrected and ortho-rectified band images that are projected onto a map, as well as geolocation information and the coordinate system used. Additionally, each image pixel provides TOA spectral reflectance data in scaled integers, conversion coefficients for spectral radiance units, viewing and solar zenith and azimuth angles, and quality flags.

The Pavia University dataset is a hyperspectral image dataset which gathered by a sensor known as the reflective optics system imaging spectrometer (ROSIS-3) over the city of Pavia, Italy. The image consists of 610×340 pixels with 115 spectral bands. The image is divided into 9 classes with a total of 42,776 labelled samples, including the asphalt, meadows, gravel, trees, metal sheet, bare soil, bitumen, brick, and shadow.

This dataset provides Level 1 radiance and Level 2 surface reflectance measured by the Airborne Visible/Infrared Imaging Spectrometer-Next Generation (AVIRIS-NG) instrument during flights over the Arctic-Boreal Vulnerability Experiment (ABoVE) domain from June to August in 2017 and July to August in 2018 and 2019. AVIRIS-NG measures reflected radiance in 425 bands at 5-nanometer (nm) intervals in the visible to shortwave infrared spectral range between 380 and 2510 nm. Measurements are radiometrically and geometrically calibrated and provided at approximately 5-meter spatial resolution. The data include 848 flight lines covering areas of interest to the ABoVE campaign over much of Alaska and western Canada. These data will allow researchers to characterize ecosystem structure and function near the height of the growing season. This dataset represents one part of a multi-sensor airborne sampling campaign conducted by eleven different aircraft teams for ABoVE. The L2 reflectance files in this publication were reprocessed with an updated reflectance algorithm and replace Versions 1 of this dataset. The imagery data are provided in ENVI format along with a RGB composite image for each flight line and shapefiles showing imagery boundaries.

Citation

In order to use these data, you must cite this data set with the following citation:

Miller, C.E., R.O. Green, D.R. Thompson, A.K. Thorpe, M. Eastwood, I.B. Mccubbin, W. Olson-Duvall, M. Bernas, C.M. Sarture, S. Nolte, L.M. Rios, M.A. Hernandez, B.D. Bue, and S.R. Lundeen. 2022. ABoVE: Hyperspectral Imagery AVIRIS-NG, Alaskan and Canadian Arctic, 2017-2019 V2. ORNL DAAC, Oak Ridge, Tennessee, USA. https://doi.org/10.3334/ORNLDAAC/2009

This dataset is openly shared, without restriction, in accordance with the EOSDIS Data Use Policy.

The study area locates in Shenzhou City, Hebei Province, China, and it consists of two sub-regions, Majiakou Village's plots(MJK_N, MJK_S) and Xijingmeng Valliage's plots(XJM). The hyperspectral data obtained on September 18th, 2019, with an electric hexacopter. The UAV carried the sensor Pika L hyperspectral imager made by Resonon company, which has a spectral range of 385nm to 1024nm with a total of 200 bands. And the flight height was set to be 100 m, resulting in a spatial resolution of 0.1 m per pixel. Spectronon and ENVI software was utilized for the post-processing of the hyperspectral data, including radiometric calibration, geometric correction, image stitching, and atmospheric correction. The images cropped to the 96*96*200 patches, then split into the Training set and Test set with a ratio of 8:2. The patches in the dataset are all numpy data type.

This dataset comprises co-aligned hyperspectral and LiDAR data collected of European beech (Fagus sylvatica) forest within core protected areas of the UNESCO Rhӧn Biosphere Reserve, Germany. Data was collected using the Headwall Hyperspec Nano sensor flown from a unmanned aerial vehicle (UAV) in September 2020. The dataset comprises image and LiDAR data of four sites, each approximately 8ha in size. The study forests were subject to the extreme drought event that impacted central Europe in 2018/2019 and this project sought to collect data to enable individual tree and stand level assessment of the response (canopy damage and defoliation) of European beech trees to extreme drought events. The hyperspectral images available in this dataset have approx. 5cm pixel size with an associated LiDAR dataset and are suitable for identifying individual trees and the degree of canopy damage (defoliation, discolouration, and mortality) sustained by individuals/stands within the forest. The work was supported by the Natural Environment Research Council (Grant NE/V00929X/1).

This folder contains a subset of hyperspectral images from the "HyperVein: A Dataset for Human Vein Detection from Hyperspectral Images" dataset, which originally consists of 200 images.

This folder includes:

1. A hyperspectral image dataset containing 10 hyperspectral images representing the left and right hand image capture of 5 volunteer participants. The images named with 'a' (e.g. 1a.bil) are left-hand images and those named with 'b' (e.g. 1b.bil) are right-hand images. Both the image file in band interleaved format (.bil) and the header (.hdr) file have been provided for each image.

2. A dataset containing the ground truth for each of the hyperspectral images. The ground truths are of dimensions 1024 by 1024 representing the selected ROI used for the vein detection experiment. This region can be mapped out using dimensions: row = 291:1314, column = 360:1383. A folder containing the ground truth images scaled to 0 and 255 for visualization purposes is also included.

3. An example MATLAB file for reading and displaying the hyperspectral images named 'Read_HS_Image'. Make sure to download and install the Hyperspectral Image Processing Toolbox on MATLAB first before running codes.

Description

ICVL is a hyperspectral image dataset, collected by "Sparse Recovery of Hyperspectral Signal from Natural RGB Images" The database images were acquired using a Specim PS Kappa DX4 hyperspectral camera and a rotary stage for spatial scanning. At this time it contains 200 images and will continue to grow progressively. Images were collected at 1392 $\times$ 1300 spatial resolution over 519 spectral bands (400-1,000nm at roughly 1.25nm increments). The .raw files contain… See the full description on the dataset page: https://huggingface.co/datasets/danaroth/icvl.

Indian Pines is a Hyperspectral image segmentation dataset. The input data consists of hyperspectral bands over a single landscape in Indiana, US, (Indian Pines data set) with 145×145 pixels. For each pixel, the data set contains 220 spectral reflectance bands which represent different portions of the electromagnetic spectrum in the wavelength range 0.4−2.5⋅10−6.

Acquisition dates: 07.07.2021; 16.07.2021; 12.09.2021
Location: Maišiagala, Vilnius District Municipality, Lithuania
Spatial resolution: 0.023 m/pixel
Number of spectral bands: 204
Spectral range: 426-958 nm (visible-near infrared spectrum)
Spectral resolution: 2.8 nm
Flight altitude: 70 m

The dataset consists of blueberry hyperspectral imaging data acquired with a UAV and a BaySpec OCI-F Hyperspectral Imager on several dates. In total, six flights on three different dates were performed. The data from each UAV flight are given as a separate dataset. Each dataset consists of raw and processed hyperspectral imaging data. The raw data include calibration images of white reference and dark background, raw hyperspectral images, and information on the UAV flight path. Calibration data are stored in the folders "...-White", "...-White_FS2", "...-Dark", and "...-Dark_FS2". Raw images are located in subfolders RawImages and RawImages_FS2 of the main data folder, which ends with "..._BI08". The BaySpec Cube Creator 2100 software was used to process raw images into hyperspectral data cubes, which are provided in the format of band sequential image files (BSQ). BSQ files are located in the Cube folders of each dataset together with HDR files containing metadata for each cube. The values of hyperspectral data cubes are in digital numbers, which can be recalculated to reflectance using the reflectance scaling factor. It is specified in the HDR files for each cube individually.

Datasheet of the dataset: https://drive.google.com/file/d/1QV5he5bGazAlN8A5Mpyl1xMc7lQcvy9W

Download links:
https://blueberry-dataset.s3.eu-west-1.amazonaws.com/2021-07-07.zip (60.48 GB)
https://blueberry-dataset.s3.eu-west-1.amazonaws.com/2021-07-16.zip (177.22 GB)
https://blueberry-dataset.s3.eu-west-1.amazonaws.com/2021-09-15.zip (81.12 GB)

The image dataset which gathered by a sensor known as the reflective optics system imaging spectrometer (ROSIS-3) over the city of Pavia, Italy.

The image consists of 610×340 pixels with 115 spectral bands. The image is divided into 9 classes with a total of 42,776 labelled samples, including the asphalt, meadows, gravel, trees, metal sheet, bare soil, bitumen, brick, and shadow

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F15506837%2F6afa9efe3e0af62e55a0032dbc9e9b34%2FScreenshot%202024-06-24%20145443.png?generation=1719221099682974&alt=media" alt="">