Comparative Analysis of CNN And SVM Machine Learning Technique for Maize Leaf Disease Detection

Publication Details

Publication Date

10/10/2026

Volume / Issue

Vol 1, Issue 5 (2026)

Article No.

044

Journal

British Journal of Contemporary Research

DOI

https://doi.org/10.67693/BJCR-LCEKL6UM

Received

20 Sep 2026

Views

37

Downloads

11

Abstract

Maize (Zea mays L.) is one of the most crucial staple crops in the world and has high susceptibility to foliar diseases like Northern Leaf Blight, Common Rust and Gray Leaf Spot that may lower the yield by as much as 40 per cent unless controlled immediately. Manual-based disease-detection methods are tedious, arbitrary and poorly sensitive to disease outbreaks, especially in resource strained agricultural environments. Another possible solution is the machine-learning methods that have a potential to automatize the diagnosis of the maize leaf diseases by using image analysis and, thus, increase diagnostic quality. The present paper made a comparative assessment of two machine-learning-based language models including the Convolutional Neural Network and Support Vector Machines in identifying maize leaf disease automatically. The standard performance indicators were used to evaluate the models and they were precision, recall, F1-score, and accuracy. According to experimental results, the CNN has a better performance of an accuracy of 95.8, macro- and weighted-average precision of 0.958-0.959, a recall of 0.958, and F1-score of 0.958, compared to the SVM which has an accuracy of 93.5 and F1-score of 0.934. The constant superiority of the CNN proves the efficiency of the learning of discriminative spatial and texture features directly in the raw images. Furthermore, the results are consistent with the available literature on plant disease detection that documents high CNN accuracy that is usually 2-6% kernel points higher than that of SVM.

License

This article is published under the Creative Commons Attribution 4.0 International License . Free to read, share, and adapt with attribution.