Research master thesis | Archaeology (research) (MA/MSc)
open access
This thesis explores the application of Deep Learning techniques for automated feature detection within GIS maps in the context of digital archaeology. Specifically, it focuses on leveraging the...Show moreThis thesis explores the application of Deep Learning techniques for automated feature detection within GIS maps in the context of digital archaeology. Specifically, it focuses on leveraging the YOLOv8s algorithm to automate the detection of prehistoric granaries on archaeological excavation maps. Traditional manual analysis methods in archaeological research are often time-consuming and labour-intensive, particularly when dealing with large spatial datasets. Moreover, the overall convoluted nature of archaeological excavations and the diverse range of features they contain present significant challenges for traditional methods. To address these challenges, this research investigates the potential of Deep Learning algorithms to enhance the efficiency and accuracy of automated feature detection on archaeological GIS maps. The results of this study demonstrate the effectiveness and potential of Deep Learning algorithms to accurately identify prehistoric granaries within archaeological excavation maps. The analysis reveals that the algorithm is able to detect and classify prehistoric granaries with a relative high degree of precision. Despite these promising results, the study underscores the challenges associated with the opacity of DL models, particularly regarding their interpretability and biases. The thesis highlights the importance of addressing issues such as data imbalance, background noise, and the inclusion of contextual information to improve the accuracy and reliability of automated detection models. While the current model demonstrates potential, further research is needed to refine these methodologies, ensuring they contribute meaningfully to archaeological analysis. This work tries to lay some foundation for future advancements in the field, advocating for the development of more comprehensive DL models that can enhance the efficiency and depth of archaeological investigations.Show less
