Systematic Baseline Evaluation Of Statistical Texture Descriptors And Lightweight Convolutional Networks For Medicinal Leaf Recognition
DOI:
https://doi.org/10.29284/y21j5586Keywords:
Medicinal plant classification, Leaf image analysis, Gray-Level Co-occurrence Matrix (GLCM), Support Vector Machine (SVM), MobileNetV2, Transfer learning, Texture feature extraction, Deep learning, Image preprocessing, Principal Component Analysis (PCA), Confusion matrix analysis, Silhouette scoreAbstract
Accurate identification of medicinal plant species is critical for applications in healthcare, pharmacology, and biodiversity conservation. Manual identification based on visual inspection of leaf characteristics is labor-intensive, subjective, and prone to error, thereby motivating the development of automated image-based classification systems. This paper presents an experimental implementation and comparative evaluation of two baseline approaches for medicinal plant leaf classification: a traditional texture-based method using Gray-Level Co-occurrence Matrix (GLCM) features combined with a Support Vector Machine (SVM) classifier, and a deep learning-based approach employing MobileNetV2 with transfer learning. In the proposed framework, leaf images undergo systematic preprocessing, including resizing, noise reduction, and morphological background removal, prior to feature extraction or deep feature learning. The GLCM-based approach extracts statistical texture descriptors such as contrast, homogeneity, energy, correlation, and angular second moment, whereas MobileNetV2 automatically learns hierarchical representations from image data. Both models are evaluated using a structured multi-class dataset consisting of medicinal plant leaf images, and performance is assessed using accuracy, F1-score, confusion matrices, silhouette score, and Principal Component Analysis (PCA) for feature space visualization. Experimental results demonstrate that the GLCM + SVM model achieves moderate classification performance, with an overall accuracy of approximately 59%, and exhibits class-specific confusion among visually similar plant species. In contrast, the MobileNetV2-based model achieves 100% classification accuracy on the evaluated dataset, as reflected by a perfectly diagonal confusion matrix, indicating strong class separability and robustness under the given experimental conditions. However, the GLCM + SVM approach offers greater interpretability and lower computational complexity, while MobileNetV2 requires higher computational resources and provides limited transparency in decision-making. The comparative analysis highlights that although deep learning models outperform traditional texture-based methods in terms of accuracy and robustness, both approaches exhibit inherent limitations related to scalability, interpretability, and generalization. These findings emphasize the necessity for a novel hybrid or enhanced classification framework that integrates the strengths of classical texture analysis and deep learning to achieve more reliable and universally applicable medicinal plant classification.
References
[1] S. Naeem, A. Ali, C. Chesneau, M. H. Tahir, F. Jamal, R. A. K. Sherwani, and M. U. Hassan, “The classification of medicinal plant leaves based on multispectral and texture features using machine learning approach,” Agronomy, vol. 11, no. 2, p. 263, 2021, doi:10.3390/agronomy11020263.
[2] J. F. Restrepo-Arias, J. M. Saldarriaga-Córdoba, and J. A. López-Hurtado, “Plant disease detection strategy based on image texture analysis,” Agriculture, vol. 12, no. 11, p. 1964, 2022, doi:10.3390/agriculture12111964.
[3] R. Demilie, A. Getahun, and T. Belay, “Plant disease detection and classification techniques: A review,” J. Big Data Anal., vol. 5, no. 1, pp. 1–20, 2024, doi:10.1186/s40537-023-00863-9.
[4] K. P. Ferentinos, “Deep learning models for plant disease detection and diagnosis,” Comput. Electron. Agric., vol. 145, pp. 311–318, 2018, doi:10.1016/j.compag.2018.01.009.
[5] S. Mohanty, D. Hughes, and M. Salathé, “Using deep learning for image-based plant disease detection,” Front. Plant Sci., vol. 7, p. 1419, 2016, doi:10.3389/fpls.2016.01419.
[6] A. Krizhevsky, I. Sutskever, and G. Hinton, “ImageNet classification with deep convolutional neural networks,” Commun. ACM, vol. 60, no. 6, pp. 84–90, 2017, doi:10.1145/3065386.
[7] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L. Chen, “MobileNetV2: Inverted residuals and linear bottlenecks,” in Proc. IEEE CVPR, 2018, pp. 4510–4520, doi:10.1109/CVPR.2018.00474.
[8] G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks,” in Proc. IEEE CVPR, 2017, pp. 4700–4708, doi:10.1109/CVPR.2017.243.
[9] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE CVPR, 2016, pp. 770–778, doi:10.1109/CVPR.2016.90.
[10] C. Szegedy et al., “Inception-v4, Inception-ResNet and the impact of residual connections on learning,” in Proc. AAAI, 2017.
[11] A. Howard et al., “MobileNets: Efficient convolutional neural networks for mobile vision applications,” arXiv:1704.04861, 2017.
[12] B. Bharadwaj, A. Mishra, and S. Bharadwaj, “Transfer learning-based CNN models for plant species identification using leaf venation patterns,” arXiv:2509.03729, 2025.
[13] S. Naeem et al., “Medicinal plant leaf classification using multi-feature extraction and SVM,” Pattern Recognit. Image Anal., vol. 31, no. 4, pp. 820–832, 2021.
[14] P. Pydipati, T. F. Burks, and W. S. Lee, “Identification of citrus disease using color texture features and discriminant analysis,” Comput. Electron. Agric., vol. 52, no. 1–2, pp. 49–59, 2006.
[15] R. Gonzalez and R. Woods, Digital Image Processing, 4th ed. Pearson, 2018.
[16] C. Cortes and V. Vapnik, “Support-vector networks,” Mach. Learn., vol. 20, no. 3, pp. 273–297, 1995.
[17] B. Scholkopf and A. Smola, Learning with Kernels, MIT Press, 2002.
[18] T. Ojala, M. Pietikäinen, and T. Mäenpää, “Multiresolution gray-scale and rotation invariant texture classification with local binary patterns,” IEEE TPAMI, vol. 24, no. 7, pp. 971–987, 2002.
[19] J. Yao, “Deep learning for plant identification and disease classification,” ACM Trans. Comput. Healthcare, 2024, doi:10.1145/3639816.
[20] V. Gautam, “A lightweight deep learning method for medicinal leaf image classification,” Sci. Rep., 2025.
[21] A. Howard et al., “Searching for MobileNetV3,” in Proc. IEEE ICCV, 2019.
[22] M. Tan and Q. Le, “EfficientNet: Rethinking model scaling for convolutional neural networks,” in Proc. ICML, 2019.
[23] S. Woo, J. Park, J.-Y. Lee, and I. S. Kweon, “CBAM: Convolutional block attention module,” in Proc. ECCV, 2018.
[24] R. Selvaraju et al., “Grad-CAM: Visual explanations from deep networks via gradient-based localization,” in Proc. IEEE ICCV, 2017.
[25] D. Chicco and G. Jurman, “The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy,” BMC Genomics, vol. 21, p. 6, 2020.
[26] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016.
[27] F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011.
[28] M. Abadi et al., “TensorFlow: Large-scale machine learning on heterogeneous systems,” 2016.
[29] D. Lowe, “Distinctive image features from scale-invariant keypoints,” Int. J. Comput. Vis., vol. 60, no. 2, pp. 91–110, 2004.
[30] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv:1409.1556, 2014.
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
