Forecasting diseases that affect plant leaves and moisture levels in the soil using a data mining approach

Abstract

The foundation of the global and Indian economies is agriculture. Since agriculture started millions of years ago, many environments, civilizations, and technical developments have fostered and defined the evolution of agricultural technology. In this study, we examine how we may analyze images of plants and soil to better keep tabs on their health, as well as how we can determine how much water each kind of plant needs. Images of the plants and soil are first taken using a digital camera with the necessary resolution. The form and geometric characteristics are extracted from the plant images using the inner distance shape context-based descriptor and geometrical descriptors. The soil images are also used to extract features and color properties. The botanical plant species dictionary is used to identify the plant type using the contour elements of the plant photos. Gradient structured random forest (GS-RF) classification is used to forecast leaf diseases. Principal Component Analysis (PCA) and Hierarchical Gradient Deep Neural Network (HG-DNN) classification techniques are used to determine the causes of a given plant disease based on the characteristics of soil images and plant disease images. The findings are communicated to the growers through text messages sent to their mobile phones on a daily and seasonal basis, along with any potential recommendations for preventative actions.

Plant phenotyping relevance

植物画像から葉の形状・幾何特性を抽出し、画像に基づく葉病害の分類・予測手法を開発しているため、植物状態の取得・推定が中心です。

Code and data availability

The article describes plant/soil image analysis (GS-RF, PCA, HG-DNN) but contains no data availability statement, no public dataset or image deposit, and no code/model availability language or URL. Only generic library mentions (Keras, Theano, TensorFlow) and the journal site URL appear; no paper-specific qualifying资产.

No evidence-backed public reproduction asset is currently recorded.