Research Brief: Improving Automated Strawberry Monitoring Through Color Calibration
June 29, 2026 
Photo: Adobe Stock
A new pilot study from University of East Anglia researchers, in collaboration with the agri-tech firm Antobot, examined color variation observed by camera-captured strawberry images and sought to determine whether color differences in ripe strawberries were biological or introduced by imaging conditions and camera software. It found that the amount of color variation in strawberry images, compared to their observable colors, may lead to inaccurate decision-making by vision-based robotics.
The study highlights the importance of computer vision systems in CEA, as they can provide data used for automated crop monitoring, ripeness assessment, pest and disease detection, and yield estimation, and how commercial CEA can achieve color consistency in imaging.
How Calibration Significantly Reduced Color Variability
The team of researchers collected images of strawberries growing in commercial polytunnels using Antobot’s robotic imaging platform. Then, a ColorChecker Color Rendition Chart was placed in each image to help researchers calibrate white balance and color correction before processing.
From 48 images featuring 112 ripe strawberries, calibration helped reduce shot-to-shot chromaticity variation by 48%. Additionally, researchers found that strawberries photographed under varying sunlight, cloud coverage, and shading conditions appeared much more consistent after calibration.
These results suggest that a significant amount of color variability in imaging systems originates from the camera technology employed, rather than from measuring methods.
The Study’s Implications on Automated Crop Monitoring and Harvesting
For commercial greenhouse and indoor farm operations, the study’s findings reinforce the importance of image quality used in crop monitoring systems. More reliable color data could improve ripeness detection, reducing training requirements for machine-learning models and enhancing decision-making based on visual crop assessments.
Additionally, researchers note that the case made for calibration is more valuable in cases where the diagnostic color signal is more subtle and unlikely to be recognized by cameras, such as when affected by nutrient deficiencies, disease symptoms, and other plant stressors.
For additional information on the study methods and findings relating to crop color calibration, please read the full paper “Colour Calibration for Agricultural Imaging: A Pilot Study” here.