ARCHAIC · Adaptive Risk Cultural Heritage Assessment In Conservation

The current project, ARCHAIC (Adaptive Risk Cultural Heritage Assessment In Conservation), aims to provide a new intelligent methodfor the risk management of heritages. In ARCHAIC, I plan to recognize physical deterioration and degradation evolution inmonuments by computer vision methods (using a combination of Neural Radiance Fields (NeRF) and social media imagery) and usethem as inputs to assess the degree of vulnerability via the neuro-fuzzy method. To compare social media photos captured at varioustimes for detecting changes, deterioration, and achieving deterioration growth, I encounter the challenge that the images vary incamera. By reconstructing 3D models of monuments at various times via NeRF and comparing images, we can effectively identifychanges and deterioration over time. Camera pose estimation is performed using Inerf, a deep neural network-based methodspecifically designed for accurate camera pose estimation of social media images. The estimated poses are utilized and given to theNeRF models to generate pair comparison viewpoints for image analysis. Finally, I consider detected deterioration’s type, boundary,amount, and growth for the fuzzy rules. ARCHAIC leads to determining the amount of intervention to manage the risks of a querycase of a monument.