Akademik Veri Yönetim Sistemi
ADVANCES IN ARCHAEOLOGICAL PRACTICE, 2026 (AHCI, Scopus)
The manual identification of ancient agricultural terraces is time-consuming and subjective, limiting large-scale archaeological landscape documentation. This study applies deep learning to detect ancient terraces in the Bozburun Peninsula, southwestern Turkey, a historically significant Hellenistic landscape. Four U-Net-based architectures were implemented-early, intermediate, and late fusion, along with an RGB-only baseline-integrating high-resolution aerial imagery (30 cm) and digital elevation models (DEMs) across 193 km2. Sixteen manually digitized areas (37.8 ha) produced 256 training patches (512 & times; 512 px). The early fusion model that combined spectral and topographic data achieved the best performance (IoU = 0.754; accuracy = 85.9%). Monte Carlo evaluation confirmed its robustness. Spatial analysis showed that 89.8% of detected terraces lie below 300 m elevation, mainly on 10 degrees-20 degrees slopes with north-northwest orientation, in agreement with previous archaeological observations. Compared with expert digitization, the model yielded higher precision (87.4% vs. 79.3%), while experts achieved higher recall (94.3% vs. 76.6%). Applied to the full peninsula, the model mapped 2,517 ha of terraces. Validation using an existing archaeological dataset (Demirciler 2014) enabled direct comparison between automated and expert-based interpretations. The results indicate the potential of deep learning for terrace detection in Mediterranean landscapes and outline a methodological framework for documenting threatened cultural heritage. Manuel olarak antik tar & imath;m teraslar & imath;n & imath; belirlemek zaman al & imath;c & imath; ve & ouml;zneldir, bu da b & uuml;y & uuml;k & ouml;l & ccedil;ekli arkeolojik peyzaj dok & uuml;mantasyonunu k & imath;s & imath;tlamaktad & imath;r. Bu & ccedil;al & imath;& scedil;ma, tarihsel olarak & ouml;nemli bir Helenistik peyzaj olan g & uuml;neybat & imath; T & uuml;rkiye'deki Bozburun Yar & imath;madas & imath;'ndaki antik teraslar & imath; tespit etmek i & ccedil;in derin & ouml;& gbreve;renmeyi uygulamaktad & imath;r. Y & uuml;ksek & ccedil;& ouml;z & uuml;n & uuml;rl & uuml;kl & uuml; hava foto & gbreve;raf & imath; (30 cm) ve say & imath;sal y & uuml;kseklik modellerini 193 km2 alan genelinde entegre eden d & ouml;rt U-Net tabanl & imath; mimari (erken, ara ve ge & ccedil; f & uuml;zyon ile sadece RGB i & ccedil;eren bir temel & ccedil;izgi) uyguland & imath;. Manuel olarak say & imath;salla & scedil;t & imath;r & imath;lm & imath;& scedil; on alt & imath; alan (37.8 ha), 256 e & gbreve;itim yamas & imath; (512 & times; 512 piksel) & uuml;retti. Spektral ve topografik verileri birle & scedil;tiren erken f & uuml;zyon modeli en iyi performans & imath; elde etti (IoU = 0.754; do & gbreve;ruluk = 85.9%). Monte Carlo de & gbreve;erlendirmesi, modelin sa & gbreve;laml & imath;& gbreve;& imath;n & imath; do & gbreve;rulad & imath;. & Ouml;nceki arkeolojik g & ouml;zlemlerle uyumlu olacak & scedil;ekilde, uzamsal analiz tespit edilen teraslar & imath;n %89.8'inin 300 metrenin alt & imath;nda bir y & uuml;kseklikte, esas olarak kuzey-kuzeybat & imath; y & ouml;nelimli 10 degrees-20 degrees e & gbreve;imlerde bulundu & gbreve;unu g & ouml;sterdi. Uzman say & imath;salla & scedil;t & imath;rmas & imath;yla kar & scedil;& imath;la & scedil;t & imath;r & imath;ld & imath;& gbreve;& imath;nda, model daha y & uuml;ksek kesinlik (Precision) sa & gbreve;lad & imath; (%87.4'e kar & scedil;& imath;l & imath;k %79.3), buna kar & scedil;& imath;n uzmanlar daha y & uuml;ksek geri & ccedil;a & gbreve;& imath;rma (Recall) elde etti (%94.3'e kar & scedil;& imath;l & imath;k %76.6). T & uuml;m yar & imath;madaya uyguland & imath;& gbreve;& imath;nda, model 2.517 ha teras alan & imath; haritalad & imath;. Mevcut bir arkeolojik veri k & uuml;mesi kullan & imath;larak yap & imath;lan do & gbreve;rulama (Demirciler 2014), otomatik ve uzman tabanl & imath; yorumlar aras & imath;nda do & gbreve;rudan kar & scedil;& imath;la & scedil;t & imath;rma yap & imath;lmas & imath;n & imath; sa & gbreve;lad & imath;. Sonu & ccedil;lar, derin & ouml;& gbreve;renmenin Akdeniz peyzajlar & imath;ndaki teras tespiti i & ccedil;in potansiyelini g & ouml;stermekte ve tehdit alt & imath;ndaki k & uuml;lt & uuml;rel miras & imath;n belgelenmesi i & ccedil;in metodolojik bir & ccedil;er & ccedil;eve sunmaktad & imath;r.