IMPLEMENTASI HYBRID PREPROCESSING MENGGUNAKAN CLAHE DAN DEEP TRANSFER LEARNING EFFICIENTNET UNTUK PENINGKATAN AKURASI KLASIFIKASI CITRA REPTIL-AMFIBI
Keywords:
CLAHE, EfficientNet, transfer learning, klasifikasi citra, reptil–amfibi.Abstract
Klasifikasi citra reptil dan amfibi merupakan tantangan signifikan dalam bidang visi komputer karena karakteristik visual objek yang kompleks, seperti pola kamuflase alami, tekstur kulit yang halus, serta kondisi pencahayaan lingkungan yang tidak konsisten. Tantangan tersebut menurunkan kemampuan model deep learning dalam mengenali fitur penting, terutama ketika dataset berukuran kecil dan tidak seimbang. Penelitian ini bertujuan meningkatkan akurasi klasifikasi citra reptil–amfibi melalui penerapan hybrid preprocessing yang menggabungkan metode Contrast Limited Adaptive Histogram Equalization (CLAHE) dan arsitektur EfficientNet berbasis transfer learning. Metode penelitian meliputi pengumpulan data, exploratory data analysis (EDA), penerapan CLAHE untuk peningkatan kontras lokal, augmentasi citra menggunakan transformasi geometrik dan fotometrik, serta pembangunan model EfficientNetB0 dengan pendekatan pelatihan dua tahap: head training dan fine-tuning. Evaluasi model dilakukan menggunakan akurasi, precision, recall, F1-score, serta analisis confusion matrix untuk membandingkan performa antara pipeline baseline dan pipeline hybrid. Hasil eksperimen menunjukkan bahwa CLAHE mampu meningkatkan keterbacaan fitur lokal sehingga model dapat mengenali tekstur reptil–amfibi secara lebih efektif. Model EfficientNetB0 dengan hybrid preprocessing menghasilkan akurasi validasi akhir sebesar 83%, lebih tinggi dibandingkan baseline yang tidak menggunakan peningkatan kontras. Peningkatan akurasi terutama terlihat pada kelas dengan tekstur jelas seperti Frog dan Chameleon, meskipun performa masih rendah pada kelas tertentu seperti Snake yang memiliki kemiripan morfologi tinggi. Penelitian ini menyimpulkan bahwa integrasi CLAHE dan EfficientNet efektif meningkatkan kualitas fitur serta robustnes model pada kondisi pencahayaan tidak seragam. Temuan ini memiliki implikasi penting bagi pengembangan sistem identifikasi satwa liar, aplikasi konservasi digital, serta penelitian lanjutan yang memanfaatkan dataset terbatas.
References
Ahmed, K., Gad, M., & …. (2024). Snake species classification using deep learning techniques.
Akhtar, N., Saeed, &, & A. (2022). Deep learning in wildlife image classification: A systematic review.
Alfano, D., P., Pastore, P., V., Rosasco, L., Odone, &, & F. (2022). Top-Tuning: A Study on Transfer Learning for an Efficient Alternative to Fine Tuning.
Alruwaili, & M. (2025). An integrated deep learning model with EfficientNet and preprocessing for classification of complex biomedical images. https://doi.org/10.1038/s41598-025-10517-w
Alzubaidi, L., Al-Amidie, M., Al-Asadi, A., Humaidi, J., A., Al-Shamma, O., Fadhel, A., M., Zhang, J., Santamaría, Duan, &, & Y. (2023). A survey on deep learning tools dealing with data scarcity: Definitions, challenges, solutions, tips, and applications.
Alzubaidi, L., Zhang, J., Humaidi, J., A., Al‐Shamma, O., Santamaría, Farina, &, & J. (2021). Review of deep learning: concepts, CNN architectures, challenges, applications, future directions.
Buriboev, S., A., Khashimov, A., Abduvaitov, Jeon, &, & H., S. (2024). CNN-based kidney segmentation using a modified CLAHE algorithm.
Cai, J., Pan, R., Lin, Liu, Zhang, L., Wen, X., Chen, Zhang, &, & X. (2023). Improved EfficientNet for corn disease identification.
Cuvelier, D., Zurowietz, M., Nattkemper, &, & T., W. (2024). Deep learning–assisted biodiversity assessment in deep-sea benthic megafauna communities: A case study in the context of polymetallic nodule mining.
Dubey, & A. (2021). Data augmentation strategies for deep learning in image classification: A review.
Halloum, K., Ez-Zahraouy, &, & H. (2025). Enhancing medical image classification through transfer learning and CLAHE optimization.
Iguernane, & M. (2025). Deep learning-based snake species identification using EfficientNet architecture.
Kumar, S., S. V, Kondaveeti, &, & H., K. (2024). Towards transparency in AI: Explainable bird species image classification for ecological research.
Lee, S. H., Park, J. M., Ko, Y. S., & Kim, C. H. (2023). Contrast enhancement-based preprocessing process to improve deep learning object task performance and results. Applied Sciences, 13(19), 10760. https://doi.org/10.3390/app131910760
Lee, Y., Chen, S., A., Tajwar, F., Kumar, A., Yao, H., Finn, &, & C. (2022). Surgical Fine-Tuning Improves Adaptation to Distribution Shifts. arXiv. https://doi.org/10.48550/arXiv.2210.11466
Lu, J., Tan, L., Jiang, &, & H. (2021). Review on convolutional neural network (CNN) applied to plant leaf disease classification.
Manole, I., & …. (2024). Enhancing dermatological diagnostics with EfficientNet: A deep learning approach.
Mittal, M., Goyal, M., L., Kaur, S., Verma, &, & A. (2023). Improving convolutional neural networks for limited datasets through enhanced preprocessing pipelines.
Mulero-Pázmány, M., Hurtado, S., Barba-González, C., Antequera-Gómez, L., M., Díaz-Ruiz, F., Real, R., Navas-Delgado, I., Aldana-Montes, &, & J., F. (2025). Addressing significant challenges for animal detection in camera trap images: a novel deep-learning-based approach.
Murcia-Gómez, D., Rojas-Valenzuela, I., & Valenzuela, O. (2022). Impact of image preprocessing methods and deep learning models for classifying histopathological breast cancer images. Applied Sciences, 12(22), 11375. https://doi.org/10.3390/app122211375
Nguyen-Tat, & B., T. (2025). Evaluating pre-processing and deep learning methods in medical image classification: benchmarking pipelines across enhancement, augmentation and transfer learning.
Nguyen, H., Q., Dao, M., Tran, &, & M. (2024). Transfer learning with EfficientNet architectures for fine-grained species recognition.
Nguyen, T. T. T., Li, W., Ogunbona, P. O., & others. (2024). A small-sized animal wild image dataset with annotations: Challenges and benchmark capacity for deep learning on reptiles, amphibians and other cryptic species. Multimedia Tools and Applications. https://doi.org/10.1007/s11042-023-16673-3
Oloko-Oba, M., Viriri, &, & S. (2021). Ensemble of EfficientNets for the diagnosis of tuberculosis.
Piffer, & S. (2024). Tackling the small data problem in medical image classification with artificial intelligence: A systematic review.
Takaya, & K. (2023). Individual identification of endangered amphibians using EfficientNetV2 and convolutional neural networks.
Vajrangi, S., Selar, Bidargaddi, A., Hiremath, R., Yeli, &, & V. (2025). Harnessing effectiveness of ResNet-50 and EfficientNet for few-shot learning.
Wen, Y., Zhang, Wang, X., Xu, &, & D. (2021). Data augmentation for deep learning: A comprehensive survey.
Wu, & Z. (2024). Image data augmentation techniques based on deep learning: A survey.
Yang, S., Xiao, W., Zhang, M., Guo, S., & Zhao, J. (2022). Image data augmentation for deep learning: A survey. https://arxiv.org/ (preprint)
Downloads
Published
Issue
Section
Citation Check
License
Copyright (c) 2025 Muchamad Iqbal Arrazzaki, Dian Ade Kurnia, Yudhistira Arie Wijaya, Edi Tohidi, Edi Wahyudin
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
