AGIcam Dataset: In-Field IoT Sensor Data for Wheat Phenotyping and Yield Prediction

Description:

AGIcam is an affordable, solar-powered Raspberry Pi-based IoT sensor system designed for high-frequency, plot-level data acquisition. Data from the AGIcam system were used to develop time-series models of wheat yield using machine learning (Random Forest) and deep learning (Long Short-Term Memory, LSTM). This dataset provides the raw and preprocessed inputs used in those analyses.

The dataset accompanies the manuscript “AGIcam: An Open-Source IoT-Based Camera System for Automated In-Field Phenotyping and Yield Prediction.” The dataset includes synchronized vegetation index (VI) metrics, weather data, and related agronomic trait data, along with sample images collected with the AGIcam system during spring and winter wheat field trials in the 2022 growing season. 

Dataset Content:

1. Vegetation Index, Weather, and Agronomic Trait Data (Spring_wheat_data.csv and Winter_wheat_data.csv)
   The AGIcam dataset includes synchronized vegetation indices, weather measurements, and agronomic trait data. Each row in the dataset represents a single VI measurement extracted from AGIcam images, paired with weather parameters recorded at the corresponding time point. Plot-level heading date and grain yield (kg/ha) are included for downstream phenotyping and modeling applications.
   
   
   • Information on column names in the CSV files
     • date: Date of data acquisition
     • timepoint: Hour of image capture
     • wheat: Wheat type (spring/winter)
     • sensor: AGIcam sensor ID
     • variety: Genotype identifier
     • rep_var: Variety replicate number
     • rep_pic: Image replicate number
     • vi: Vegetation index name
     • max: Maximum VI value
     • mean: Mean VI value
     • median: Median VI value
     • std: Standard deviation of VI values
     • p95 / p90 / p85: 95th, 90th, and 85th percentile VI values
     • avg_air_temp: Average air temperature (°C)
     • humidity: Relative humidity (%)
     • avg_soil_temp_8_in: Soil temperature at 8-inch depth (°C)
     • precip: Precipitation (mm)
     • solar_rad: Solar radiation (W/m²)
     • heading_date: Date of heading stage
     • yield_kg/ha: Grain yield (kg per hectare)
       
       
2. Sample Images (AGIcam8.zip and AGIcam17.zip)
   Selected RGB and NoIR images captured from AGIcam sensors to demonstrate image quality and time resolution.

Data Collection and Processing:

• Sensors: AGIcam units (RGB and NoIR cameras)
• Weather: ATMOS 41 sensor station 
• Location: Spring and winter wheat field trials in Pullman, Washington State.
• Image Frequency: 3 captures per day per sensor.

Use Case: This dataset can be used to:

• High-frequency VI and weather data modeling
• Time-series analysis of crop growth
• Machine/deep learning yield prediction
• Sensor fusion in breeding programs

Acknowledgments:
This study was funded by the United States Department of Agriculture (USDA) - National Institute for Food and Agriculture (NIFA) competitive project (accession number 1028108), hatch project (accession number 1014919), and Washington State University’s College of Agricultural, Human, and Natural Resource Sciences’ Emerging Research Issues competitive grant opportunity (ERI-20-04). The authors would like to thank Dr. Milton Valencia Ortiz and Kingsley Charles Umani for their support during the field camera installation

Contact Information:
For further inquiries regarding this dataset, please contact Sindhuja Sankaran (s.sankaran@wsu.edu).