MODEL EFISIEN BERBASIS MOBILE VISION TRANSFORMER (MOBILEVIT) UNTUK KLASIFIKASI JENIS TANAH DARI CITRA

Darmatasia; M. Hasrul H

doi:10.24252/instek.v10i2.61891

Authors

Darmatasia Universitas Islam Negeri Alauddin Makassar
M. Hasrul H Universitas Islam Negeri Alauddin Makassar

DOI:

https://doi.org/10.24252/instek.v10i2.61891

Keywords:

Jenis Tanah, Mobile Vision Transformer, Model Efisien

Abstract

Identifikasi jenis tanah berperan krusial dalam sektor pertanian. Namun, metode klasifikasi konvensional seperti uji laboratorium dan observasi langsung masih memiliki keterbatasan dalam hal efisiensi waktu, biaya, dan skala. Penelitian ini mengusulkan model efisien untuk klasifikasi citra tanah dengan arsitektur Mobile Vision Transformer. Pendekatan transfer learning digunakan dalam membangun model untuk mengatasi keterbatasan jumlah data latih yang selanjutnya disesuaikan dengan data jenis tanah melalui fine-tuning. Dataset yang digunakan dalam penelitian ini diperoleh dari platform Kaggle, yang terdiri dari enam kelas yaitu Arid Soil, Black Soil, Laterite Soil, Mountain Soil, Red Soil, dan Yellow Soil. Dataset dibagi menjadi 80% untuk pelatihan dan 20% untuk pengujian, dengan evaluasi kinerja berdasarkan akurasi, presisi, recall, F1-score, dan waktu inferensi. Eksperimen dilakukan dengan membandingkan performa MobileViT terhadap model konvensional seperti CNN ringan (MobileNet) dan Vision Transformer standar. Hasil penelitian menunjukkan bahwa model yang diusulkan mampu mencapai keseimbangan antara akurasi dan efisiensi komputasi, dengan tingkat akurasi sebesar 97%. MobileViT menunjukkan efisiensi waktu inferensi yang lebih baik dibandingkan Vision Transformer standar, dengan kecepatan sekitar 5 kali lebih cepat. Kecepatan inferensi MobileViT mendukung penerapannya pada aplikasi real-time berbasis perangkat dengan daya komputasi terbatas tanpa menurunkan akurasi dengan signifikan. Meskipun demikian, penelitian ini memiliki keterbatasan pada ukuran dataset yang terbatas.

Downloads

Download data is not yet available.

References

[1] D. N. Kiran Pandiri, R. Murugan, and T. Goel, “Smart soil image classification system using lightweight convolutional neural network,” Expert Syst. Appl., vol. 238, p. 122185, 2024, doi: https://doi.org/10.1016/j.eswa.2023.122185.

[2] M. Javaid, A. Haleem, I. H. Khan, and R. Suman, “Understanding the potential applications of Artificial Intelligence in Agriculture Sector,” Adv. Agrochem, vol. 2, no. 1, pp. 15–30, 2023, doi: https://doi.org/10.1016/j.aac.2022.10.001.

[3] S. Alugoju and P. P, “A Smart Agricultural Framework for Soil Image Classification Using Modified DenseNet and Crop Recommendation System Using Random Forest,” Int. J. Eng. Trends Technol., vol. 72, pp. 119–129, Oct. 2024, doi: 10.14445/22315381/IJETT-V72I10P112.

[4] P. J. and S. T., “Deep learning based multi-labelled soil classification and empirical estimation toward sustainable agriculture,” Eng. Appl. Artif. Intell., vol. 119, p. 105690, 2023, doi: https://doi.org/10.1016/j.engappai.2022.105690.

[5] Y. Aydin, U. Isikdag, G. Bekdas, S. M. Nigdeli, and Z. W. Geem, “Use of Machine Learning Techniques in Soil Classification,” Sustainability, vol. 15, no. 3. 2023, doi: 10.3390/su15032374.

[6] M. D. Nguyen et al., “Novel approach for soil classification using machine learning methods,” Bull. Eng. Geol. Environ., vol. 81, no. 11, p. 468, 2022.

[7] N. Barkataki, S. Mazumdar, P. B. D. Singha, J. Kumari, B. Tiru, and U. Sarma, “Classification of soil types from GPR B Scans using deep learning techniques,” in 2021 International Conference on Recent Trends on Electronics, Information, Communication & Technology (RTEICT), 2021, pp. 840–844, doi: 10.1109/RTEICT52294.2021.9573702.

[8] E. K. Gyasi and S. Purushotham, “Advancements in soil classification: an in-depth analysis of current deep learning techniques and emerging trends,” Air, Soil Water Res., vol. 16, p. 11786221231214068, 2023.

[9] Y. li, H. Liu, and T. Lv, “A multi-task learning model for global soil moisture prediction based on adaptive weight allocation,” Sci. Rep., vol. 15, no. 1, p. 18631, 2025, doi: 10.1038/s41598-025-01894-3.

[10] R. K. Banoth and B. V. R. Murthy, “Soil image classification using transfer learning approach: MobileNetV2 with CNN,” SN Comput. Sci., vol. 5, no. 1, p. 199, 2024.

[11] D. Sivabalaselvamani, L. Rahunathan, K. Nanthini, T. Harshini, and C. Hariprasath, “Soil Classification using Deep Learning Techniques,” in 2023 Second International Conference on Augmented Intelligence and Sustainable Systems (ICAISS), 2023, pp. 582–586, doi: 10.1109/ICAISS58487.2023.10250710.

[12] M. Leonardi and A. Chandra, “Analisis Perbandingan CNN dan Vision Transformer untuk Klasifikasi Biji Kopi Hasil Sangrai,” J. MEDIA Inform. BUDIDARMA, vol. 8, p. 1398, Jul. 2024, doi: 10.30865/mib.v8i3.7732.

[13] V. Hassija et al., “Transformers for Vision: A Survey on Innovative Methods for Computer Vision,” IEEE Access, vol. PP, p. 1, Jan. 2025, doi: 10.1109/ACCESS.2025.3571735.

[14] F. Sheth, “Comprehensive Soil Classification Datasets,” 2025. .

[15] F. Sheth, P. Mathur, A. K. Gupta, and S. Chaurasia, “An advanced artificial intelligence framework integrating ensembled convolutional neural networks and Vision Transformers for precise soil classification with adaptive fuzzy logic-based crop recommendations,” Eng. Appl. Artif. Intell., vol. 158, p. 111425, 2025.

[16] S. Mehta and M. Rastegari, “Mobilevit: light-weight, general-purpose, and mobile-friendly vision transformer,” arXiv Prepr. arXiv2110.02178, 2021.

[17] A. Dosovitskiy et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv Prepr. arXiv2010.11929, 2020.

[18] G. Y. Gopal and M. A. Amer, “Mobile vision transformer-based visual object tracking,” arXiv Prepr. arXiv2309.05829, 2023.

[19] Z. Somogyi, “Performance evaluation of machine learning models,” in The Application of Artificial Intelligence: Step-by-Step Guide from Beginner to Expert, Springer, 2021, pp. 87–112.

[20] A. Howard et al., “Searching for MobileNetV3,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 1314–1324, doi: 10.1109/ICCV.2019.00140.

[21] M. G. Lanjewar and O. L. Gurav, “Convolutional Neural Networks based classifications of soil images,” Multimed. Tools Appl., vol. 81, no. 7, pp. 10313–10336, 2022, doi: 10.1007/s11042-022-12200-y.