Analysis of Automatic Segmentation Ability using Segment Anything Model (SAM) of Rural Settlement Objects in Indonesia on Orthophoto Images
Abstract
On-screen digitization is a time-consuming and labor-intensive process that limits the efficiency of mapping, particularly when analyzing rural settlements with numerous objects. This study investigates the potential of the Segment Anything Model (SAM) to improve both accuracy and efficiency in automatic segmentation of rural settlements in Indonesia using orthophotos. SAM was applied to two regions representing different settlement distribution patterns, and its performance was assessed with Intersect over Union (IoU) metrics at both regional and building levels. Results show that SAM achieved strong overall accuracy, with IoU values ranging from 0.883 to 0.891. At the building level, Langgudu and Tilongkabila districts showed consistently high accuracy. Crucially, SAM reduced on-screen digitization time by 60% to 86,26%, underscoring its ability to enhance efficiency without being constrained by settlement distribution patterns. These findings highlight SAM’s potential to streamline large-scale mapping workflows, and further research with larger datasets and broader geographic coverage is recommended to validate its generalizability and strengthen its role in geospatial analysis and planning.
References
[1] N. Bashit, Y. Prasetyo, and A. Sukmono, “Kajian Perkembangan Lahan Terbangun Kota Pekalongan Menggunakan Metode Urban Index (Ui),” Elipsoida J. Geod. dan Geomatika, vol. 2, no. 02, pp. 12–18, 2019, doi: 10.14710/elipsoida.2019.6440.
[2] R. Saidi et al., “Pemanfaatan Citra Landsat 8 untuk Identifikasi Sebaran Vegetasi di Kecamatan Pangandaran,” JPG (Jurnal Pendidik. Geogr., vol. 7, no. 1, pp. 1–14, 2020, doi: 10.20527/jpg.v7i1.7607.
[3] D. R. Septiani, S. Subiyanto, and F. J. Amarrohman, “Identifikasi Kesesuaian Dan Intensitas Pemanfaatan Lahan Di Kelurahan Lamper Lor Menggunakan Foto Udara Tahun 2018,” J. Geod. Undip, vol. 9, no. 2, pp. 71–80, 2020, [Online]. Available: https://ejournal3.undip.ac.id/index.php/geodesi/article/view/27168
[4] A. Roihan, P. A. Sunarya, and A. S. Rafika, “Pemanfaatan Machine Learning dalam Berbagai Bidang: Review paper,” IJCIT (Indonesian J. Comput. Inf. Technol., vol. 5, no. 1, pp. 75–82, 2020, doi: 10.31294/ijcit.v5i1.7951.
[5] M. Farwati, I. Talitha Salsabila, K. Raihanun Navira, T. Sutabri, and U. Bina Darma Palembang, “Analisa pengaruh teknologi artificial intelligence (AI) dalam kehidupan sehari-hari [Analyze the influence of artificial intelligence (AI) technology in daily life],” J. Sist. Inform. dan Menejemen, vol. 11, no. 1, pp. 41–42, 2023.
[6] Q. Wu and L. P. Osco, “samgeo: A Python package for segmenting geospatial data with the Segment Anything Model (SAM),” J. Open Source Softw., vol. 8, no. 89, p. 5663, 2023, doi: 10.21105/joss.05663.
[7] D. Wang et al., “SAMRS: Scaling-up Remote Sensing Segmentation Dataset with Segment Anything Model,” Adv. Neural Inf. Process. Syst., vol. 36, no. NeurIPS, 2023.
[8] C. Ayu, S. Kinasih, and H. Hidayat, “Ekstraksi Data Bangunan Dari Data Citra Unmanned Aerial Vehicle Menggunakan Metode Convolutional Neural Networks ( CNN ) ( Studi Kasus : Desa Campurejo , Kabupaten Gresik ),” vol. 17, no. 1, pp. 81–92, 2021.
[9] A. N. Jamaludin, Sosiologi Perdesaan, vol. 53, no. 9. 2015.
[10] R. Sasongko, W. Astuti, and G. Yudana, “Pola Spasial Permukiman Di Bantaran Sungai Premulung, Kota Surakarta,” Desa-Kota, vol. 4, no. 2, p. 152, 2022, doi: 10.20961/desa-kota.v4i2.59526.152-166.
[11] A. Kirillov et al., “Segment Anything,” Proc. IEEE Int. Conf. Comput. Vis., pp. 3992–4003, 2023, doi: 10.1109/ICCV51070.2023.00371.
[12] M. Ilyas, “Deteksi Pelanggaran Berkendara Dengan Metode Yolo (You Only Look Once),” vol. 106, no. 8, pp. 8–25, 2020, [Online]. Available: https://elibrary.unikom.ac.id/id/eprint/4108/%0Ahttps://elibrary.unikom.ac.id/id/eprint/4108/8/UNIKOM_Muhammad Ilyas_BAB 2.pdf
[13] F. Ramasari, F. Firdaus, S. Nita, and K. Kartika, “Penggunaan Metode You Only Look Once dalam Penentu Pindah Tanaman Cabai Besar Ternotifikasi Telegram,” Elektron J. Ilm., vol. 13, no. November, pp. 45–52, 2021, doi: 10.30630/eji.13.2.229.
[14] B. Xie, H. Tang, B. Duan, D. Cai, Y. Yan, and G. Agam, “MaskSAM: Towards Auto-prompt SAM with Mask Classification for Volumetric Medical Image Segmentation,” 2025, [Online]. Available: http://arxiv.org/abs/2403.14103
[15] L. Lucchese and S. K. Mitray, “Color image segmentation: A state-of-the-art survey,” Proc. Indian Natl. Sci. Acad. (INSA-A). Delhi, Indian Natl Sci Acad, vol. 67, pp. 207–221, 2001, [Online]. Available: http://ultra.sdk.free.fr/docs/Image-Processing/filters/Color Image Segmentation-A State-of-the-Art Survey.pdf%5Cnhttp://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.84.4896