Apple tree disease detection using VGG16 & Inceptionv3

Abstract

In recent years, the term 'Precision agriculture' has gained tremendous attention due to its ability to improve crop yield while reducing its cost. This can be achieved by applying modern Machine learning and Computer vision techniques. In this paper, we propose a comprehensive study of the following deep learning models VGG16 and InceptionV3. Our approach involves training these models individually on a data-set of apple-tree leaf images, which include healthy and diseased trees,and evaluating their performance on a validation set. We also conduct experiments to compare the performance of the two models and find a distinguishable difference between their Validation accuracy. Our results show that both VGG16 and InceptionV3 achieve high accuracy in detecting apple tree diseases, with InceptionV3 outperforming VGG16 slightly. We believe that our method can be applied in real-world scenarios to help farmers detect and manage diseases in their apple trees, leading to higher crop yield and reduced costs.

Plant phenotyping relevance

リンゴ葉画像から健全・罹病状態を推定する深層学習手法が研究の中心であり、VGG16とInceptionV3の性能比較・検証も行っているため、植物病害フェノタイピング手法に該当する。

Code and data availability

The paper's phenotyping measurements (apple leaf disease classification with VGG16/InceptionV3) are based on the public New Plant Diseases Dataset from Kaggle, which the authors explicitly state is freely available with no access restrictions. No author code, models, or checkpoints are deposited.