Multispectral imaging dataset of potato plants on the field, acquired using UAV

An experiment:

• The experiment was conducted in Central Slovenia, at an elevation of 320 meters above sea level.
• The region has a subalpine climate with an average temperature of 9.5°C and average precipitation of 1352 mm.
• Eight potato varieties were planted in a randomized block design with four replications.
• Each block contained one microplot of a single potato variety.
• Each plot was a 6x3 meter section containing approximately 70 potato plants distributed into four rows.
• The experiment was conducted in two different fields, one organic and one conventional.
• Potato seeds were planted in mid-April and grown during the 2022 growing season, typically from April to September in Central Slovenia.

Multispectral imaging: 

• Multispectral images of potato field crops were captured using a SKYHERO SPYDER X4-850 GEO Edition Unmanned Aerial Vehicle (UAV) with a Micasense Rededge MX 5-band multispectral camera.
• A Stonex S9i GPS device was used alongside ground control points (GCPs) to increase the spatial accuracy of the images and for georeferencing.
• Imaging was performed around noon, with the UAV maintaining an altitude of 50 m above ground level and a flight speed of 5 m/s.
• The flight plan included an 80% bidirectional overlap and a 25% buffer to ensure data completeness and redundancy.
• After acquisition, the images underwent a standard series of processing procedures using PIX4DMapper and QGIS. This included georeferencing, orthorectification, radiometric correction, conversion to reflectance units, and the construction of mosaic representations.

Field measuring and plant health assessment :   

• Ground measurements included various parameters such as leaf chlorophyll concentration (SPAD), stomatal conductance (gsw), transpiration (E), quantum efficiency in light (PhiPS2), and electron transport rate (ETR).
• The SPAD-502 meter was used to quantify SPAD, and the Li 600 device was used to assess physiological values.
• The presence of Alternaria sp. was determined through visual assessment by qualified evaluators.
• Measurements were conducted within the central area of each plot to mitigate border effects.
• Four plants of each variety were selected, resulting in 16 measurements per variety per image and a total of 128 measurements per image across all eight varieties.
• This resulted in a total of 768 measurements (384 for ecological and 384 for conventional) across all three time points.
• Alternaria visual assessments were conducted only during the final time point, yielding 256 measurements (128 for ecological and 128 for conventional).
• GPS coordinates were recorded for all plants during the first time point, enabling direct pixel-to-measurement matching in the image. These coordinates were used for the two subsequent time points.

Files:

• File structure:

    📁 data
    ├── 📁 multispectral_images
    │   ├── 📁 2022_07_20__konv_ecobreed
    │   │   ├── 🖼️ Ecobreed_krompir_konv_20_07_2022_transparent_reflectance_nir_modified.tif
    │   │   ├── 🖼️ Ecobreed_krompir_konv_20_07_2022_transparent_reflectance_red_modified.tif
    |   |   └── 🖼️ ...
    │   ├── 📁 2022_07_20__eko_ecobreed
    │   │   ├── 🖼️ Ecobreed_krompir_eko_20_07_2022_transparent_reflectance_nir_modified.tif
    │   │   ├── 🖼️ Ecobreed_krompir_eko_20_07_2022_transparent_reflectance_red_modified.tif
    |   |   └── 🖼️ ...
    │   └── 📁 ...
    ├── 📁 shapefiles
    │   ├── 📄 potato_locations_eko.shp
    │   ├── 📄 potato_locations_konv.shp
    │   ├── 📄 potato_measured_locations_eko.shp
    │   ├── 📄 potato_measured_locations_konv.shp
    │   └── 📄 ...
    └── 📁 measurements
        ├── 📊 Alternaria_ocenjevanje1_Ecobreed_krompir_2022.xlsx
        ├── 📊 LICOR_Ecobreed_krompir_2022.xlsx
        ├── 📊 SPAD_Ecobreed_krompir_2022.xlsx
        └── 📊 Varieties_grouped_Ecobreed_krompir_2022.xlsx

          multispectral_images: This directory contains raster images captured by a multispectral camera. The images are organized into subdirectories by date and treatment type (conventional-konv or ecological-eko). Each image file, such as Ecobreed_krompir_konv_20_07_2022_transparent_reflectance_nir_modified.tif, represents a specific spectral band i.e., near-infrared of the multispectral image.

           shapefiles: This directory contains shapefiles (.shp), which are used to store the geographic coordinates of potato plants. Files like potato_locations_eko.shp and potato_locations_konv.shp contain the locations of all potato plants for ecological and conventional treatments, respectively. Meanwhile, files like potato_measured_locations_eko.shp and potato_measured_locations_konv.shp contain the locations of potato plants where physiological measurements were taken.

           measurements: This directory contains Excel (.xlsx) files with various ground measurements taken from the potato plants. Each file represents a different type of measurement, such as Alternaria_ocenjevanje1_Ecobreed_krompir_2022.xlsx for Alternaria scoring, LICOR_Ecobreed_krompir_2022.xlsx for LICOR measurements, SPAD_Ecobreed_krompir_2022.xlsx for SPAD measurements, and Varieties_grouped_Ecobreed_krompir_2022.xlsx for grouped variety data.

Additional links:

Code where the dataset was used for the entire analysis could be found here:

Manuscripts-code/Potato-UAV-exploratory--CEA-2024: Field-Scale UAV-Based Multispectral Phenomics: Leveraging Machine Learning, Explainable AI, and Hybrid Feature Engi-neering for Enhancements in potato phenotyping

Paper could be found here:

Field-scale UAV-based multispectral phenomics: Leveraging machine learning, explainable AI, and hybrid feature engineering for enhancements in potato phenotyping - ScienceDirect