20.500.12556/RUL-189598 Uporaba globokega učenja za prepoznavanje arheoloških značilnosti na visoko ločljivih digitalnih modelih višin Application of deep learning methods for the recognition of archaeological features on high-resolution digital elevation models Aerolasersko skeniranje (ALS) je v zadnjih dveh desetletjih bistveno razširilo možnosti arheološke prospekcije, hkrati pa je eksponentno povečalo količino podatkov, ki jih je treba pregledati, kar odpira prostor za metode globokega učenja pri (delni) avtomatizaciji prepoznave arheoloških struktur. Magistrska naloga preverja zmogljivost odprtokodnega orodja ADAF (Automatic Detection Of Archaeological Features) pri samodejni zaznavi gomil, ene najpogostejših, a hkrati morfološko izjemno heterogenih arheoloških struktur v slovenskem prostoru. Ker je bil privzeti ADAF model naučen na irskih in angleških gomilah, geomorfološko in kulturno različnih od slovenskih, je bila najprej izvedena ročna interpretacija desetih izbranih slovenskih gomilnih najdišč iz različnih obdobij in regij, pri čemer je bilo evidentiranih 374 gomil, nato pa so bili na lokalno pridobljenih podatkih (10.294 objektov, od tega 942 gomil) naučeni trije novi modeli, katerih rezultati so bili primerjani s privzetimi. Analiza je pokazala, da privzeti model prepozna le 7, 75 % ročno evidentiranih gomil, medtem ko doučenje z lokalnimi podatki uspešnost skoraj podvoji (najboljši novi model doseže 24, 3-odstotno prekrivnost), pri čemer je ključni morfološki dejavnik uspešne prepoznave zaobljenost gomile, sledi pa ji definiranost njenih robov, medtem ko naklon in površina gomile nanjo vplivata zanemarljivo. Na podlagi rezultatov se sklepa, da ADAF v trenutni obliki deluje predvsem kot orodje za začetno preiskavo območij z morfološko dobro ohranjenimi gomilami, in ne kot nadomestilo za strokovno terensko interpretacijo, za bistveno izboljšanje prenosljivosti in zanesljivosti modelov pa bo v prihodnje potrebna kakovostna razširitev in terenska verifikacija učnih množic ter sistematično testiranje prenosljivosti modelov med različnimi regijami in kulturnimi konteksti. Airborne laser scanning (ALS) has, over the past two decades, substantially expanded the possibilities of archaeological prospection, while at the same time exponentially increasing the volume of data that needs to be reviewed, opening up space for deep learning methods in the (partial) automation of archaeological feature recognition. This master's thesis tests the performance of the open-source tool ADAF (Automatic Detection of Archaeological Features) in the automatic detection of barrows (gomile), one of the most common yet morphologically extremely heterogeneous archaeological structures in the Slovenian territory. Since the default ADAF model was trained on Irish and English barrows, which are geomorphologically and culturally different from Slovenian ones, a manual interpretation of ten selected Slovenian barrow sites from different periods and regions was first carried out, in which 374 barrows were recorded; three new models were then trained on locally acquired data (10,294 objects, of which 942 were barrows), and their results were compared with those of the default model. The analysis showed that the default model detects only 7.75% of the manually recorded barrows, whereas fine-tuning with local data nearly doubles the success rate (the best new model achieves 24.3% overlap), with barrow roundness being the key morphological factor for successful detection, followed by the definition of its edges, while slope and area of the barrow have a negligible effect. Based on the results, it is concluded that ADAF, in its current form, functions primarily as a tool for the initial survey of areas with morphologically well-preserved barrows, rather than as a substitute for expert field interpretation; substantially improving the transferability and reliability of the models will, in the future, require a quality expansion and field verification of the training datasets, as well as systematic testing of model transferability across different regions and cultural contexts. ADAF samodejna prepoznava gomile globoko učenje ALS- vizualizacije ADAF automatic detection barrows deep learning ALS visualizations true false false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2026-10-10 07:16:16 2026-10-10 07:16:23 2026-10-12 14:14:35 0000-00-00 00:00:00 2026 0 Ljubljana 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 643448 74790.pdf 74790.pdf 1 2DFA88A0160DBDC8BF27A8124BDE50F4 8b75dd40e8ccb641868a9b1d964c29fece578ba4f4bda38791ac64ab3db233cc 95ff5cbc-c469-11f1-8bc5-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=250262 Filozofska fakulteta 0 0 0