PlantVillage and PlantDoc dataset mediated plant disease predictions: A perspectives review

Abstract

Plant disease analysis is crucial for the better yield of the crops, and correct detection of particular infection on the plant parts provide the basis of better control of the plant disease. The prediction and control of disease in crop plants is essential for the food security. The technological advancements, particularly, in the field of the artificial intelligence and machine learning have provided impetus for newer dimensions of application of technology in different fields including the plant crop disease. The fundamental database of different infections in different crop plants forms the basis of the standard training of the machine learning algorithms which further predicts the disease on the test samples. The more detailed dataset of plant diseases with corresponding large number of sample examples helps in better training of the machine learning (ML) modules. The collections of disease dataset by the PlantVillage and evaluated PlantDoc dataset are being extensively used for the ML training and prediction of disease. This perspective discussion delves in the fundamental different types of plant diseases of the different parts of plant (leaf, fruits, flowers, stem), particularly of the crop plants, with emphasis on PlantVillage and PlantDoc datasets. The evaluation of ML techniques for conclusive detection of the disease possibilities has further been included in the discussion.

Plant phenotyping relevance

植物病害の画像データセットと機械学習による植物病害検出を中心に扱うレビューであり、植物の病害状態を観測・推定するフェノタイピング手法に該当する。

Code and data availability

This is a perspectives review of PlantVillage and PlantDoc datasets with no authors' own phenotyping measurements, images, code, models, or data deposits. The datasets discussed are cited prior public resources, not paper-specific assets of this article.

No evidence-backed public reproduction asset is currently recorded.