Detection of calcium deficiency in indoor-grown lettuce under LED lighting using computer vision

Abstract

Calcium deficiency and its associated physiological disorders, such as tip burn, pose considerable challenges for indoor hydroponic lettuce production, impacting both yield and quality. This study presents a novel approach combining advanced image segmentation and classification techniques to detect calcium deficiency in lettuce during its growth stages under colored LED lighting. Early detection of nutrient deficiencies is crucial for timely intervention and efficient nutrient management. This experiment involved growing butterhead lettuce plants under a Deep-Water Culture (DWC) system with controlled calcium treatments. Preprocessing techniques, including Contrast Limited Adaptive Histogram Equalization (CLAHE) and Red Color Correction (RCC), were applied to enhance quality and consistency and augmented to improve model generalization. Our methodology employed a two-stage process that allowed us to leverage the strengths of specialized models. First, lettuce leaves were segmented from the background using state-of-the-art models, including U-Net, U-Net++, Recurrent U-Net, and Inception U-Net. U-Net++ demonstrated the highest segmentation accuracy (98.56 %) with robust generalization compared to the other models. Segmentation isolates the region of interest and removes background noise, enabling the classifier to focus more effectively on disease-related features. Deep learning classification models, such as ResNet and EfficientNet, were applied in the second stage to detect calcium deficiency from the segmented images. EfficientNetB2 emerged as the most reliable classifier, achieving an accuracy of 91.51 % on the RCC dataset, while Resnet50 achieved a comparable accuracy of 91.18 % on the same dataset. This study highlights the potential of integrating deep learning models into automated hydroponic systems for real-time nutrient monitoring, offering a practical solution to enhance productivity and sustainability in indoor hydroponic farming.