ANALISIS PENGARUH URUTAN PREPROCESSING TERHADAP KINERJA MOBILENETV2 DAN VGG16 UNTUK KLASIFIKASI PENYAKIT DAUN CABAI

Syafa Nabila Putri Samsuri; Martanto; Yudhistira Arie Wijaya; Dodi Solihudin; Tati Suprapti

Authors

Syafa Nabila Putri Samsuri STMIK IKMI CIREBON, Indonesia
Martanto STMIK IKMI Cirebon, Indonesia
Yudhistira Arie Wijaya STMIK IKMI Cirebon, Indonesia
Dodi Solihudin STMIK IKMI Cirebon, Indonesia
Tati Suprapti STMIK IKMI Cirebon, Indonesia

Abstract

This study examines the impact of different preprocessing sequences resizing, normalization, and augmentation on the performance of two Convolutional Neural Network (CNN) architectures, MobileNetV2 and VGG16, for classifying chili leaf disease, specifically Anthracnose. A dataset of 870 images consisting of two classes (Healthy and Anthracnose) was used to evaluate three preprocessing combinations: Resizing → Normalization → Augmentation (R→N→A), Resizing → Augmentation → Normalization (R→A→N), and Normalization → Augmentation → Resizing (N→A→R). The experimental results show that MobileNetV2 achieves its highest accuracy of 99.31% with the R→A→N sequence, whereas VGG16 performs best with the R→N→A sequence, reaching an accuracy of 95.14%. These findings indicate that preprocessing order significantly influences model performance and is not universally applicable across CNN architectures. The study highlights the importance of aligning preprocessing pipelines with model characteristics to optimize classification accuracy in plant disease detection.

References

Algoritma space vector. (2020). Jurnal Amplifier, 5(2), 1–6.

Ashurov, A. Y., Al Gaashani, M. S. A. M., Samee, N. A., Alkanhel, R., Atteia, G., Abdallah, H. A., & Muthanna, M. S. A. (2025). Enhancing plant disease detection through deep learning: A depthwise CNN with squeeze and excitation integration and residual skip connections. Frontiers in Plant Science, 15, 1505857. https://doi.org/10.3389/fpls.2024.1505857

Atila, U., & Uçar, A. (2021). Plant disease classification using deep transfer learning. Expert Systems with Applications, 168, 114363.

Ferentinos, E. (2022). Deep learning models for plant disease detection and diagnosis. Computers and Electronics in Agriculture, 145, 311–318.

Gomez, H., & Pineda, R. (2023). Evaluation of convolutional architectures for crop disease identification. Journal of Imaging, 9(1), 1–10.

Le, A. T., Nguyen, H. M., & Tran, D. H. (2024). Optimizing plant disease classification with hybrid convolutional recurrent neural networks and image preprocessing. Applied Sciences, 14(19), 9118. https://doi.org/10.3390/app14199118

Rahman, S. (2023). Systematic evaluation of augmentation and preprocessing pipelines in deep learning-based plant disease recognition. Artificial Intelligence in Agriculture, 6, 23–34.

Wagner, K. (2021). Transfer learning for agricultural image classification: A comprehensive review. IEEE Access, 9, 112233–112250.

Wang, X., Chen, L., & Zhao, Y. (2022). Deep learning-based detection of plant leaf diseases under complex field conditions. Computers and Electronics in Agriculture, 198, 1–12.

Xu, Q., Liu, H., & Li, Z. (2021). Impact of preprocessing strategies on CNN performance for image-based crop disease classification. Information Processing in Agriculture, 8(4), 572–584.