Reconstruction of Historical Buildings Using Social Media Data: A Case Study from the World Cultural Heritage List

Tunalıoğlu Öcalan N., Erdoğan B., Öcalan T., Arıcan D., Altuntaş C., Arda T., ...Daha Fazla

INTERNATIONAL JOURNAL OF ENGINEERING AND GEOSCIENCES, cilt.11, sa.1, ss.21-39, 2026 (ESCI)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 11 Sayı: 1
  • Basım Tarihi: 2026
  • Doi Numarası: 10.26833/ijeg.1635501
  • Dergi Adı: INTERNATIONAL JOURNAL OF ENGINEERING AND GEOSCIENCES
  • Derginin Tarandığı İndeksler: Emerging Sources Citation Index (ESCI), Central & Eastern European Academic Source (CEEAS), Directory of Open Access Journals
  • Sayfa Sayıları: ss.21-39
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

Cultural and natural heritage assets, as a set of tangible and intangible values that reveal the shared past and historical accumulations of the communities living together, are important not only for the past but also for the transfer to future generations and are the responsibility of all nations of the world. Today's digitalized information age, development, and change in science and technology contribute to producing highly accurate three-dimensional (3D) inventories of these cultural heritages. However, in addition to natural degradation and destruction processes, unexpected events such as war, terrorist attacks, and natural disasters can hinder the formulation of the traditional inventories. At this point, it is important to carry out 3D modeling studies using crowdsourced images and videos from social media, to enhance modeling accuracy and support digital documentation, virtual museum initiatives, and heritage preservation. This study aims to present a strategy following the Structure-fromMotion approach to create 3D models of cultural heritage assets by using shared crowdsourced images and videos collected via social media platforms and applying a normalization procedure for scale standardization to assess model accuracy. Data obtained during nighttime, snowy conditions, or overly filtered captures were excluded, and a segmentation procedure using the Segment Anything model was implemented to remove irrelevant objects. As a result, an approach that can be used in inventory studies has been presented by using images and videos shared by users on social media platforms through an integrated and mutually supportive methodology. Quantitative analysis using the M3C2 method showed that RMSE values ranged from 0.0010 to 0.0036 across the models, with over 93% of the matched points falling within ±1σ. These results confirm the reliability of the proposed approach for large-scale digital heritage documentation.