Tracing Past Agrarian Field Systems Through Ai-Based Analysis of Satellite Imagery in Konya, Central Anatolia, Türkiye

KÜÇÜKDEMİRCİ M.

Archaeological Prospection, cilt.32, sa.4, ss.881-897, 2025 (SCI-Expanded, AHCI, Scopus) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 32 Sayı: 4
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1002/arp.2001
  • Dergi Adı: Archaeological Prospection
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Arts and Humanities Citation Index (AHCI), Scopus, Academic Search Premier, IBZ Online, Periodicals Index Online, Anthropological Literature, Art Source
  • Sayfa Sayıları: ss.881-897
  • Anahtar Kelimeler: agricultural field systems, convolutional neural networks, deep learning, Google Earth satellite imagery, support vector machines
  • İstanbul Üniversitesi-Cerrahpaşa Adresli: Evet

Özet

This study investigates the use of Artificial Intelligence (AI) techniques for the automated classification and detection of historical agrarian fields in Konya, Central Anatolia, Türkiye. Three training strategies were evaluated: (1) a standalone support vector machine (SVM), (2) fine-tuning of pretrained convolutional neural networks (CNNs) and (3) a hybrid approach that uses pretrained models as feature extractors in combination with an SVM for classification. Among the three approaches tested, fine-tuning the ResNet50V2 model emerged as the most flexible and computationally efficient method. Tests conducted on seven satellite image patches demonstrated high classification accuracy and notable model robustness. Despite these promising results, the limited availability of very high-resolution satellite imagery constrained model's broader applicability to larger test areas. Overall, the findings underscore the potential of AI-based models for historical land-use determination, offering archaeologists a valuable tool for both pre-survey targeting and postsurvey validation.