Home / Current Issue / Paper 1717169
Machine Learning and Deep Learning for Plant Disease Classification and Detection
Subject area: Science,Engineering and Technology · Area of research: Artificial Intelligence
DOI: https://doi.org/10.64388/IREV9I10-1717169
Abstract
Plant diseases cause significant losses in global agricultural output, and early, accurate identification remains a major challenge. This paper presents a unified deep learning pipeline that simultaneously localises and classifies plant diseases from leaf images. A YOLOv5s model performs region-level detection, while a transfer-learning-based ResNet50 classifier identifies the specific disease category. The system is trained on the PlantVillage dataset, which contains 54,306 images spanning 38 disease classes across 14 crop species. Preprocessing includes image resizing, normalisation, and a combination of standard and advanced augmentation strategies, namely CutMix and MixUp. ResNet50 is fine-tuned in two phases with Cosine Learning Rate decay, and Test Time Augmentation (TTA) is applied at inference for additional accuracy gains. The ResNet50 classifier achieves 99.1% validation accuracy with TTA, while the YOLOv5s detection model reaches a mAP@0.50 of 0.856. The macro-averaged precision, recall, and F1-score stand at 0.974, 0.968, and 0.971, respectively. The integrated pipeline demonstrates reliable end-to-end performance with 97.6% accuracy on 200 held-out images, offering a practical and scalable solution for precision agriculture.
Keywords
Plant Disease Detection, Deep Learning, ResNet50, YOLOv5, Transfer Learning, Precision Agriculture, CutMix, MixUp, Test Time Augmentation, PlantVillage.
How to cite this paper
@article{1717169,
author = {N. Sri Janaki Ram, N. Venkata Sai Gopi Chandu, S. Ajay, Reddy Veeramohanarao},
title = {Machine Learning and Deep Learning for Plant Disease Classification and Detection},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {10},
pages = {3382-3385},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717169.pdf},
abstract = {Plant diseases cause significant losses in global agricultural output, and early, accurate identification remains a major challenge. This paper presents a unified deep learning pipeline that simultaneously localises and classifies plant diseases from leaf images. A YOLOv5s model performs region-level detection, while a transfer-learning-based ResNet50 classifier identifies the specific disease category. The system is trained on the PlantVillage dataset, which contains 54,306 images spanning 38 disease classes across 14 crop species. Preprocessing includes image resizing, normalisation, and a combination of standard and advanced augmentation strategies, namely CutMix and MixUp. ResNet50 is fine-tuned in two phases with Cosine Learning Rate decay, and Test Time Augmentation (TTA) is applied at inference for additional accuracy gains. The ResNet50 classifier achieves 99.1% validation accuracy with TTA, while the YOLOv5s detection model reaches a mAP@0.50 of 0.856. The macro-averaged precision, recall, and F1-score stand at 0.974, 0.968, and 0.971, respectively. The integrated pipeline demonstrates reliable end-to-end performance with 97.6% accuracy on 200 held-out images, offering a practical and scalable solution for precision agriculture.},
keywords = {Plant Disease Detection, Deep Learning, ResNet50, YOLOv5, Transfer Learning, Precision Agriculture, CutMix, MixUp, Test Time Augmentation, PlantVillage.},
month = {April},
doi = {https://doi.org/10.64388/IREV9I10-1717169}
}