Integrating Deep Learning and mSVM for Category of Class Prediction of Spice Plant Leaf Diseases

Abstract

Abstract E arly disease detection is fundamental to protect the crops and provide early treatment. Plants with medicinal values are rare and need complete care. Black pepper is a spice herb, highly used as medicinal plant. The diseases prominent in black pepper spice are anthracnose, phytophthora, slow wilt, quick wilt, and yellowing. The proposed work suggests deep learning-based prediction and classification. The proposed work utilizes a benchmark dataset created in real time environment. The dataset is preprocessed, segmented, and labeled into classes under expert supervision. Deep neural network ResNet − 50 is trained with novice data. The trained features are saved in an array. This data is matched in linear with extracted feature of Support vector machine SVM, a machine learning algorithm used for disease prediction. This is done by passing and mapping the hyperparameter features of Renet-50 as an array and mapping it to feature array of multiclass support vector machine. Hence there is no requirement of training machine learning separately. This results in fast training and early disease prediction .

Plant phenotyping relevance

黒コショウ葉の病害症状を画像から分類・予測する深層学習／SVM手法と、前処理・セグメンテーション・ラベリング済みデータセットを中心に扱っており、植物病態の画像ベース表現型計測に該当する。

Code and data availability

The paper describes a self-collected black pepper leaf image dataset (1300 images) and a ResNet-50 + mSVM pipeline, but contains no public dataset deposit, no code availability statement, no repository URL, and no supplement reference. The dataset is described as 'created in real time environment' with no sharing or de

No evidence-backed public reproduction asset is currently recorded.