In the realm of cultural heritage preservation, a groundbreaking study led by A. El-Alaily from the 3D Survey Group at the Politecnico di Milano is pushing the boundaries of 3D digitization. The research, published in ‘The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences’ (translated to English as ‘International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences’), combines Visual-SLAM (Simultaneous Localization and Mapping) and semantic segmentation to create more interpretable 3D models of historical sites. This innovation could have significant implications for the energy sector, particularly in the preservation and maintenance of heritage sites that are integral to cultural tourism and local economies.
Traditional methods of 3D semantic enrichment have often fallen short when it comes to identifying small-scale geometric elements or visually ambiguous classes. El-Alaily’s study addresses this limitation by leveraging the rich contextual and textural information of 2D imagery. “We use deep learning-based 2D semantic segmentation techniques to detect challenging semantic categories, such as fine architectural elements and material decay,” El-Alaily explains. These detections are then projected into 3D space through a 2D-to-3D semantic segmentation framework that couples V-SLAM and 3D results with the 2D predictions.
The framework was evaluated using data acquired with the fish-eye multi-camera mobile mapping system ATOM-ANT3D in two challenging case study environments. The results demonstrate a reliable level of accuracy, enhancing the interpretability of 3D models by providing meaningful and metrically interpreted object classifications.
For the energy sector, this research could be a game-changer. Heritage sites often require careful maintenance and restoration, which can be both costly and time-consuming. By providing more accurate and detailed 3D models, this technology can help identify areas of decay or structural weakness, allowing for more targeted and efficient maintenance. This could lead to significant cost savings and reduced downtime for sites that are crucial to local economies.
Moreover, the ability to create more interpretable 3D models could also enhance the visitor experience. Tourists could interact with detailed, semantically enriched models, gaining a deeper understanding and appreciation of the sites they visit. This could drive more tourism, benefiting local businesses and the economy as a whole.
El-Alaily’s research is a significant step forward in the field of cultural heritage preservation. By combining V-SLAM and semantic segmentation, it offers a more efficient and accurate way to document and interpret 3D data. As the technology continues to evolve, it could have far-reaching implications for the energy sector and beyond, shaping the future of heritage preservation and tourism.