Journals / İstanbul Sabahattin Zaim Üniversitesi Fen Bilimleri Enstitüsü Dergisi / 2026 / Cilt: 8 Sayı: 1

Ai-Powered Building Detection and Automated 3d Digital Twin Generation From Satellite Imagery

Yapay Zeka Destekli Bina Tespiti ve Uydu Görüntülerinden Otomatik 3d Dijital İkiz Oluşturma

Buse Eryaş(YILDIZ TEKNİK ÜNİVERSİTESİ)0009-0000-9477-6317
Pages
50–73
Publication Date
2026-06-30
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Particularly when detecting small and closely spaced building footprints, YOLO frames with local constraints offer higher accuracy than standard models (Xie et al., 2020).

📌 3 Key Takeaways

🎯
Objective & Hypothesis:The increasing demand for up-to-date and accurate geographic information due to rapid urbanization has made the development of urban digital twins a critical research priority.
📊
Key Findings & Data:Open-source platforms such as OpenStreetMap (OSM) suffer from significant data gaps, particularly in rural and rapidly developing areas.
💡
Conclusion & Impact:Particularly when detecting small and closely spaced building footprints, YOLO frames with local constraints offer higher accuracy than standard models (Xie et al., 2020).
🔬Design:Modeling & Simulation
📍Scope:2024 Verileri
⚙️Variables:Dijital İkiz · Bina Tespiti · Uydu Görüntüsü

Abstract

The increasing demand for up-to-date and accurate geographic information due to rapid urbanization has made the development of urban digital twins a critical research priority. Digital twin technology enables real-time monitoring, simulation, and analysis through virtual representations of physical assets. Open-source platforms such as OpenStreetMap (OSM) suffer from significant data gaps, particularly in rural and rapidly developing areas. To address this problem, this study proposes an end-to-end digital twin generation system that combines AI-based building detection from high-resolution satellite imagery with open-source data. Accurate determination of building boundaries and locations is a fundamental requirement in many critical areas such as security, health, environmental management, and economic planning (Tasyurek, 2024). Traditional building detection methods often encounter difficulties due to the complexity and density of objects in satellite imagery; this has necessitated the use of deep learning-based architectures (Yap et al., 2024). The proposed system uses YOLO architectures for object detection and transfers the obtained data to a 3D environment via Cesium and Unreal Engine 5 platforms. Particularly when detecting small and closely spaced building footprints, YOLO frames with local constraints offer higher accuracy than standard models (Xie et al., 2020).

Özet

Hızlı kentleşme nedeniyle güncel ve doğru coğrafi bilgilere olan artan talep, kentsel dijital ikizlerin geliştirilmesini kritik bir araştırma önceliği haline getirmiştir. Dijital ikiz teknolojisi, fiziksel varlıkların sanal temsilleri aracılığıyla gerçek zamanlı izleme, simülasyon ve analiz olanağı sağlar. Ancak OpenStreetMap (OSM) gibi açık kaynaklı platformlar, özellikle kırsal ve hızla gelişen bölgelerde önemli veri eksikliklerinden muzdariptir. Bu sorunu çözmek amacıyla bu çalışma, yüksek çözünürlüklü uydu görüntülerinden yapay zekâ tabanlı bina tespitini açık kaynaklı verilerle birleştiren uçtan uca bir dijital ikiz üretim sistemi önermektedir. Bina sınırlarının ve konumlarının doğru bir şekilde belirlenmesi, güvenlik, sağlık, çevre yönetimi ve ekonomik planlama gibi pek çok kritik alanda temel bir gerekliliktir (Tasyurek, 2024). Geleneksel bina tespit yöntemleri, uydu görüntülerindeki nesnelerin karmaşıklığı ve yoğunluğu nedeniyle sıklıkla zorluklarla karşılaşmaktadır. Bu durum, derin öğrenme tabanlı mimarilerin kullanımını zorunlu kılmıştır (Yap ve ark., 2024). Önerilen sistem, nesne tespiti için YOLO mimarilerini kullanmakta ve elde edilen verileri Cesium ve Unreal Engine 5 platformları aracılığıyla 3B ortama aktarmaktadır. Özellikle küçük ve birbirine çok yakın konumlanmış bina ayak izlerinin tespitinde, yerel kısıtlamalara sahip YOLO çerçeveleri standart modellere kıyasla daha yüksek doğruluk sağlamaktadır (Xie ve ark., 2020).

🔖

Cite This Article

Eryaş, B. (2026). Ai-Powered Building Detection and Automated 3d Digital Twin Generation From Satellite Imagery. İstanbul Sabahattin Zaim Üniversitesi Fen Bilimleri Enstitüsü Dergisi, Cilt: 8 Sayı: 1, 50–73. https://doi.org/10.47769/izufbed.1930617

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (11)

  1. Bshouty, E., & Dalyot, S. (2021). Calculating OpenStreetMap building heights from single user-generated photographs. Mapping and Geo-Information Engineering, The Technion, Haifa, Israel.
  2. Chen, J. (2019). Software development of first person virtual interactive art exhibition hall based on UE4. Western Leather, 2019(12), 24-26.
  3. Demircan, D., Palabıyık, S. (2025). Dijital İkiz Teknolojilerinin Eğitime Adaptasyonu, Potansiyelleri ve Sınırlılıkları. International Journal of Social and Humanities Sciences Research (JSHSR), 12(118): 795-810, doi: 10.5281/zenodo.15018000.
  4. Gillies, S. (n.d.). Shapely user manual. Retrieved May 8, 2026, from https://shapely.readthedocs.io/en/stable/manual.html
  5. Hidayatullah, W. G. B., Dewandaru, A., & Wicaksono, T. B. (2025). Procedural content generation with large language models for three-dimensional environment design. 2025 IEEE International Conference on Data and Software Engineering (ICoDS). IEEE.
  6. Hussain, M., Huang, J., Liu, X., Duan, Y., & Wu, H. (2026). YOLO-guided SAM for accurate building segmentation in remote sensing images. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 19, 4403–4417.
  7. Kaushal, M., & Kumar, A. (2022). Rapid-YOLO: A novel YOLO based architecture for shadow detection. Optik, 260, 169084. doi: 10.1016/j.ijleo.2022.169084
  8. Radany, N., & Abdelaziz, M. (2024). Shadow detection of building facades for energy efficiency using YOLOv8 and segmentation techniques. 2024 5th International Conference on Artificial Intelligence, Robotics and Control (AIRC), 23-28. doi: 10.1109/AIRC61399.2024.10672357
  9. Santhoshi, M., & Sneha, T. (2025). Deep neural networks for extraction of buildings from satellite images. 2025 2nd International Conference on Research Methodologies in Knowledge Management, Artificial Intelligence and Telecommunication Engineering (RMKMATE), 1-8. doi: 10.1109/RMKMATE64874.2025.11042773
  10. Tunç, H. C., Yıldırım, S. G., & Durusu, A. (2024). Artificial Intelligence-Based Estimation of Energy Production and Consumption of Buildings from Satellite and Aerial Footage. 2024 International Conference on Smart Systems and Power Management (IC2SPM), 1-6. doi: 10.1109/IC2SPM62723.2024.10841349
  11. Zhou, D., Fang, J., Song, X., Guan, C., Yin, J., Dai, Y., & Yang, R. (2019, September). IoU loss for 2D/3D object detection. In 2019 International Conference on 3D Vision (3DV) (pp. 85-94). IEEE.