Sunflower Disease detection using Ensemble Deep Learning Models

Abstract

Plant diseases, such as fungi and parasites, disrupt or alter the plant’s essential functions. In order to prevent potential economic losses from these plant diseases, early diagnosis of these diseases is crucial. The use of Deep Learning models for the detection and classification of plant diseases has been found to dramatically increase the speed of diagnosis while at the same time minimizing the amount of error involved. Taking advantage of a variety of tools and techniques, including transfer learning and ensemble learning, we have experimented with different deep-learning models and pre-existing architectures to find out which combination would best fit the data we gathered from the Sun Flower Fruits and Leaves dataset. Moreover, our best model has the ability to classify and detect sunflower disease with an accuracy of 97.91%, which is a considerable improvement over state-of-the-art models currently used for the classification and detection of sunflower disease.

Plant phenotyping relevance

ヒマワリ葉・果実画像から病害状態を分類・検出する深層学習手法を比較・評価しており、植物病害表現型の取得・推定が中心である。

Code and data availability

The paper's sunflower disease image dataset is explicitly declared publicly available on Mendeley Data with a direct URL in the Data Availability statement; no author code or models are shared.