Citation

Al Mahin Khan, Abdulla and Lee, It Ee and Chuah, Teong Chee and Wali, Qamar and Chung, Gwo Chin and Luangxaysana, Khanthanou (2026) A Systematic Evaluation of Machine Learning Algorithms and Handcrafted Features Across Multiple Plant Disease Datasets. In: 15th International Symposium on Communication Systems, Networks and Digital Signal Processing, CSNDSP 2026, 15 July 2026 - 17 July 2026, Edinburgh.

Text 3.pdf - Published Version Restricted to Repository staff only Download (558kB)

Abstract

This paper systematically analyzes classical machine learning models with handcrafted features in the classification of plant diseases in varying structural complexities in datasets. An integrated experimental pipeline was used on three datasets that corresponded to a binary, imbalanced and multi-class setting through image preprocessing, Haralick and HOG feature extraction and classification by SVM, Random Forest (RF), and KNN. 5-fold stratified cross-validation was used to assess the model's performance, and metrics like accuracy, precision, F1-score, per-class recall, and the analysis of the confusion matrix were implemented. Findings demonstrate that classical models are very accurate when applied in controlled binary data, but it becomes much less accurate when imbalanced and multi-class datasets. The early performance saturation indicated by learning curve analysis presents the problem of limitations that are not caused by lack of data but instead caused by the feature representation. Per-class analysis demonstrated that it persistently misclassifies visually similar disease groups which are not reflected in overall accuracy. These results show that handcrafted feature-based models do not have the representational ability to support credible plant disease classification in real world situations.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Deep Learning , Machine Learning
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	01 Oct 2026 02:53
Last Modified:	01 Oct 2026 02:53
URII:	http://shdl.mmu.edu.my/id/eprint/16758

Downloads

Edit (login required)