Scholarlink

<p>Salinity stress severely constrains crop productivity, creating an urgent need for rapid, scalable, and non-destructive diagnostic tools compatible with intelligent farming systems. This study presents an integrated phenotyping and analytics framework that combines RGB imaging, machine learning, and edge enabled wireless sensor network for accurate detection of three levels of salinity stress, 0 mM, 100 mM, and 200 mM of NaCl, in peppermint. Daily lateral and nadir images were acquired over a ten–day period and subjected to robust color space segmentation, after which two complementary feature extraction strategies were applied, namely RGB–based and histogram–based representations, with vegetation indices computed for both feature sets alongside morphological descriptors. Machine learning models were evaluated with emphasis on predictive accuracy, computational efficiency, and suitability for deployment on resource constrained devices. Nadir image features consistently outperformed lateral perspectives, achieving classification accuracies of up to 98.30% using a Fine KNN model with only two features. Under specific compact feature–model configurations, a Wide Neural Network, with a memory footprint of 0.010&#xa0;MB, achieved a near perfect classification (100%) on the independent test set, while also exhibiting negligible generalization gaps between validation and testing performance, suggesting stable model behavior under the evaluated experimental conditions rather than evident overfitting. Histogram–based feature analysis further achieved 85.26% accuracy under nadir configuration with a compact 7.08&#xa0;MB model. SHAP interpretability analysis revealed that salinity stress signatures were dominated by reductions in vegetation greenness, increased hue variability, and canopy contraction, possibly reflecting chlorophyll degradation and growth inhibition. This work establishes RGB–based edge intelligence as a practical and biologically meaningful solution for scalable salinity stress monitoring in smart agriculture.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Edge computing and rgb phenotyping enables machine learning‑driven salinity monitoring in peppermint

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月27日

查看全文
  • Ali Ahmad,
  • Vinie Lee Silva-Alvarado,
  • Sandra Sendra,
  • Jaime Lloret

摘要

Salinity stress severely constrains crop productivity, creating an urgent need for rapid, scalable, and non-destructive diagnostic tools compatible with intelligent farming systems. This study presents an integrated phenotyping and analytics framework that combines RGB imaging, machine learning, and edge enabled wireless sensor network for accurate detection of three levels of salinity stress, 0 mM, 100 mM, and 200 mM of NaCl, in peppermint. Daily lateral and nadir images were acquired over a ten–day period and subjected to robust color space segmentation, after which two complementary feature extraction strategies were applied, namely RGB–based and histogram–based representations, with vegetation indices computed for both feature sets alongside morphological descriptors. Machine learning models were evaluated with emphasis on predictive accuracy, computational efficiency, and suitability for deployment on resource constrained devices. Nadir image features consistently outperformed lateral perspectives, achieving classification accuracies of up to 98.30% using a Fine KNN model with only two features. Under specific compact feature–model configurations, a Wide Neural Network, with a memory footprint of 0.010 MB, achieved a near perfect classification (100%) on the independent test set, while also exhibiting negligible generalization gaps between validation and testing performance, suggesting stable model behavior under the evaluated experimental conditions rather than evident overfitting. Histogram–based feature analysis further achieved 85.26% accuracy under nadir configuration with a compact 7.08 MB model. SHAP interpretability analysis revealed that salinity stress signatures were dominated by reductions in vegetation greenness, increased hue variability, and canopy contraction, possibly reflecting chlorophyll degradation and growth inhibition. This work establishes RGB–based edge intelligence as a practical and biologically meaningful solution for scalable salinity stress monitoring in smart agriculture.