Multi-Crop Leaf Disease Detection using Deep Learning Methods

Abstract

The image processing technique is a method useful in agricultural processes for enhancing accuracy and uniformity of processes in farming while decreasing farmers’ manual observation. Leaf disease detection using deep learning applications can be helpful for farmers to analyze the affected leaves at an early stage which will in turn aid in the agricultural process. In this paper, we have used Convolution Neural Network (CNN), a deep learning algorithm mainly used for analyzing visual imagery, for the detection of various crop leaf diseases. The CNN-based model will help in differentiating between diseased and healthy leaves, which will improve farmers’ harvest quality. The main objective of the paper is to create a new dataset that contains three plant leaves that are cauliflower, tomato, and mango, and then compare the accuracy using various CNN models which are generally used for unstructured datasets i.e., images. Also analyzing the results on the basis of different experimental configurations such as choice of deep learning architecture, choice of dataset type, and Choice of training-testing set distribution. Results achieved from these experiments display the performance and precision of the model best fit for disease detection of plants.

Plant phenotyping relevance

植物葉の画像から病害状態を推定するCNN手法を開発・比較し、新規データセットも作成しているため、病害表現型の取得・推定が中心です。

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.