COMPACT CONVOLUTIONAL NEURAL NETWORK ARCHITECTURE FOR ONION DISEASE CLASSIFICATION USING CROP IMAGES

Abstract

Detecting disorders in crops at an initial phase is important for improved agricultural productivity. Different diseases like purple blotch in onions affect crop quality worldwide. Traditional approaches for identifying purple blotch require time, wide examination, and frequent farm observation. With technical improvements in recent years, agriculturalists have been able to discover optimal solutions that have caused higher harvests. This article presents compact Convolutional Neural Network (CNN) architecture for onion disease (purple blotch) classification from crop images. This Onion Crop Disease Dataset (OCDD) contains 1000 images of healthy and infected crops. Four distinct architectures InceptionV3, Xception, EfficientNetB7, and DenseNet201 are compared. Subsequent trials for assessment are applied. DenseNet201 delivers 94.54% accuracy in comparison to other models.

Plant phenotyping relevance

タマネギ画像から病害状態を分類するCNN手法の開発・比較・評価が研究の中心であり、植物病害フェノタイピングに該当する。

Code and data availability

The paper's OCDD image dataset (1,000 onion crop images) and Keras training scripts are described but no public deposit, repository, or availability URL is provided; the text even states no public onion disease dataset exists. No paper-specific public asset qualifies.

No evidence-backed public reproduction asset is currently recorded.