Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM69253
Volume:
10 |
11 |
Nov 2026
Pages:
1-8
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.

Information Science and Engineering
Jawaharlal Nehru New College of Engineering

Information Science and Engineering
Jawaharlal Nehru New College of Engineering

Information Science and Engineering
Jawaharlal Nehru New College of Engineering

Information Science and Engineering
Jawaharlal Nehru New College of Engineering

Information Science and Engineering
Jawaharlal Nehru New College of Engineering
Fungal diseases of grapevine leaves, black rot and esca (black measles) in particular, reduce yield and are often noticed late, because early lesions look alike to anyone without training. We built an image classifier for four leaf conditions (black measles, black rot, Isariopsis leaf spot and healthy) from 13,234 labelled photographs divided roughly 70/15/15 into training, validation and test sets. Three ImageNet-pretrained backbones were adapted to the task: EfficientNetB3 with a frozen base, ResNet50 trained in two stages with partial fine-tuning, and a VGG16 model with a small dense head. On the held-out test set EfficientNetB3 reached 98.69% accuracy (26 errors in 1,990 images) and ResNet50 reached 99.55% (9 errors in 1,989 images, macro F1 of 0.9955). Nearly all remaining mistakes involve black rot and black measles, and six of the nine ResNet50 errors were made with more than 99.8% confidence, which matters for anyone planning to deploy such a model. We also describe a web prototype that takes a leaf photograph and returns the predicted class, class probabilities and short management advice. Finally, we document a preprocessing mistake that held our first EfficientNetB3 run at 41.86% test accuracy, since it is an easy trap when these backbones are reused.
KEYWORDS
We compared transfer-learning classifiers for grape leaf disease on a fixed four-class dataset and built a web prototype around the trained model. ResNet50 with two-stage finetuning reached 99.55% test accuracy and EfficientNetB3 with a frozen base reached 98.69%. In both cases the hard part is separating black rot from black measles, and ResNet50 occasionally calls black rot leaves healthy. Confident mistakes and poor behaviour on cluttered photographs show that dataset accuracy alone is a weak indicator of field readiness. Next we plan to evaluate the VGG16 checkpoint on the same test folders and add it to Table III, fine-tune EfficientNetB3 with the same two-stage recipe, remove duplicates from the split, and collect field images from local vineyards. On the application side we want Grad-CAM maps [16] so users can see which part of the leaf drove a decision, a reject option for uncertain cases, and a TensorFlow Lite build for use on a phone without a network connection.
[1] “Transfer Learning for Multi-Crop Leaf Disease Image Classification using Convolutional Neural Network VGG,” Artificial Intelligence in Agriculture, vol. 6, pp. 23–33, 2022, doi: 10.1016/j.aiia.2021.12.002. [2] I. Kunduracioglu and I. Pacal, “Advancements in deep learning for accurate classification of grape leaves and diagnosis of grape diseases,” Journal of Plant Diseases and Protection, vol. 131, pp. 1061–1080, 2024, doi: 10.1007/s41348-024-00896-z. [3] “GAPNet: Single and multiplant leaf disease classification method based on simplified SqueezeNet for grape, apple and potato plants,” PeerJ Computer Science, 2025, doi: 10.7717/peerj-cs.2941. [4] “Comparative Analysis of Transfer Learning Models for the Detection and Classification of Grape Leaf Diseases,” in Proc. 8th International Conference on Parallel, Distributed and Grid Computing (PDGC), 2024, doi: 10.1109/PDGC64653.2024.10984031. [5] “Real-Time Automatic Detection of Grape Leaf Diseases with MobileNetV2 Deep Learning Network Based on Artificial Intelligence,” in Proc. Advanced Computing and Communication Technologies for High Performance Applications (ACCTHPA),2025, doi: 10.1109/ACCTHPA65749.2025.11168579. [6] X. Xie, Y. Ma, B. Liu, J. He, S. Li, and H. Wang, “A Deep-Learning-Based Real-Time Detector for Grape Leaf Diseases Using Improved Convolutional Neural Networks,” Frontiers in Plant Science, vol. 11, Art. no. 751751, 2020. [7] R. Dwivedi, S. Dey, C. Chakraborty, and S. Tiwari, “Grape Disease Detection Network Based on Multi-Task Learning and Attention Features,” IEEE Sensors Journal, vol. 21, no. 16, pp. 17573–17580, 2021. [8] S. Mousavi and G. Farahani, “A Novel Enhanced VGG16 Model to Tackle Grapevine Leaves Diseases With Automatic Method,” IEEE Access, vol. 10, pp. 113245–113260, 2022. [9] R. Karthik, R. Menaka, S. Ompirakash, P. Bala Murugan,M. Meenakashi, S. Lingaswamy, and D. Won, “GrapeLeafNet: A Dual-Track Feature Fusion Network With Inception-ResNet and Shuffle-Transformer for Accurate Grape Leaf Disease Identification,” IEEE Access, vol. 12,pp. 14567–14580, 2024. [10] L. Falaschetti, L. Manoni, R. C. F. Rivera, D. Pau, G. Romanazzi, O. Silvestroni, V. Tomaselli, and C. Turchetti, “A Low-Cost, Low-Power and Image Detector for Grape Leaf Esca Disease Based on a CNN,” IEEE Journal on Emerging and Selected Topics in Circuits and Systems, vol. 11, no. 3, pp. 468–480, 2021.
Publication Date
11/20/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
0
Total views since publication
PDF Downloads
1
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...