Comprehensive Survey of Plant Disease Detection and Classification based on Machine Learning and Deep Learning Algorithms

Abstract

Plant diseases are a major problem worldwide, affecting crop yield and quality and impacting food security. Early identification and accurate diagnosis of plant diseases is crucial to minimizing crop damage and maximizing eco-friendly farming practices. Traditional methods of plant disease detection include manual diagnosis by experts, which is time consuming to arrive at the final diagnosis and is prone to human errors. In recent years, Machine Learning (ML) and Deep Learning (DL) algorithms have been used for the automatic detect plant diseases, with no intervention needed by experts. These algorithms are used to detect diseases by analysing images of plant stems, leaves and other parts, differentiating between healthy and diseased plants. Step by step procedures are used to diagnose plant diseases by employing techniques such as image pre-processing, feature extraction, and classification to enhance image quality for accurate disease classification and prediction. This paper details the application of various existing ML and DL algorithms in the detection, classification and prediction of plant diseases, where image-based techniques are also adopted for accurate and swift diagnosis. Performance metrics such as sensitivity, accuracy, and specificity elaborate on the efficacy of ML and DL algorithms, acting as reliable tools for accurate plant disease detection. The outcomes of this study illustrate that these ML and DL models are well suited for the identification and classification of plant diseases.

Plant phenotyping relevance

植物画像から病害状態を検出・分類する機械学習手法を体系的に扱うレビューであり、植物フェノタイピング手法が中心です。

Code and data availability

This is a survey of ML/DL plant disease detection methods. It describes third-party datasets (PlantVillage, cassava, Cercospora) and reviews prior studies, but provides no authors' own phenotype datasets, images, code, models, or any availability statements or URLs. No paper-specific public assets qualify.

No evidence-backed public reproduction asset is currently recorded.