Development and Validation of a MobileNetV2 Convolutional Neural Network for Automated Diagnosis of Tomato Fungal Diseases in Northern Nigeria

Abstract

Accurate, field-deployable diagnostic tools are needed to close the diagnostic gap that limits fungicide targeting among smallholder tomato farmers in Northern Nigeria. This study developed and validated a lightweight convolutional neural network (CNN) for automated diagnosis of five major tomato fungal diseases plus three additional common conditions, trained on field-collected leaf images from Kano and Kaduna States. A MobileNetV2 architecture pre-trained on ImageNet was fine-tuned via transfer learning on more than 10,000 images across ten disease and health classes, using farm-level dataset splitting to prevent data leakage and five-fold cross-validation for model selection.

Plant phenotyping relevance

トマト葉画像から病害・健全状態を直接推定するCNNを開発し、データ分割と交差検証で技術検証しており、植物状態の取得・推定手法が中心である。

Code and data availability

The paper describes a field-collected tomato leaf image dataset (>10,000 images) and a fine-tuned MobileNetV2 model, but contains no data availability statement, repository deposit, or public URL for the dataset, images, code, or trained model/checkpoints. All URLs in the text are citations to prior work, not paper-own

No evidence-backed public reproduction asset is currently recorded.