Leaf Pathology Detection in Potato and Pepper Bell Plant using Convolutional Neural Networks

Abstract

Agriculture is the backbone of world’s economy. This sector faces predominant issues in recognizing crop infection, disease prediction, pest control, weed detection and yield prediction leading to the shortfall in both quality and production of food. To ensure food safety, high resilience and increased crop yields, the precise diagnosis and recognition of underlying plant disease along with classification of crops from weeds is vital. The recent advancements in automatic feature extraction and classification techniques using Artificial Intelligence have gained attraction in the field of agriculture and crop protection. This paper proposes a Novel Convolutional Neural Network model for crop disease classification. The model is trained and tested in publicly available Plant Village Dataset with 38 categories and 15 classes. For the experimental analysis, the model is trained with 5 classes which includes potato and pepper bell categories. Further, the performance of the proposed model is analyzed with machine leaning models such as Support Vector Machine (SVM), K-Nearest Neighborhood (K-NN), Random Forest, Decision Tree and have attained the highest accuracy of 91.28%. In the testing phase, it is observed that this model is superior in terms of accuracy, specificity, precision, recall and F1-Score.

Plant phenotyping relevance

植物画像から病害状態を分類するCNN手法の開発・性能比較が中心であり、植物病害フェノタイピングに該当する。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.