PSI - Issue 64

Francesco Calabrò et al. / Procedia Structural Integrity 64 (2024) 1759–1766 1765 Francesco Calabrò, Giovanna E. Minniti, Antonino Fotia, Raffaele Pucinotti / Structural Integrity Procedia 00 (2019) 000 – 000 7 and arrangement within the image. Its ability to discern between different types of damages, such as surface cracks or deeper lesions, has contributed to a detailed assessment of the structural condition. Furthermore, the neural network has shown good generalization, recognizing fractures under various lighting conditions and shooting angles. This robustness is crucial to ensure practical applicability in real-world situations where environmental conditions may vary. The results obtained by the CNN have been integrated into a three-dimensional model, enabling a comprehensive visualization of the lesions within the structural context. This provides operators and experts with an important visual resource to assess the distribution of fractures and gain a better understanding of the overall structural condition.

Fig. 3. A)3 D model b) crack particular on photogrammetric 3D c)3D model with crack highlighted by CNN d) particular of 3D The 3D modeling process, incorporating the latest in photogrammetric and computer vision techniques, will provide an accurate representation of the building's geometry and the spatial distribution of the identified lesion. This model will not only serve as a visual aid for understanding the extent of the damage but will also facilitate in-depth measurements, including crack dimensions, orientations, and any associated deformations. The integration of georeferencing data will further enhance the accuracy of the model, ensuring a seamless alignment with the building's real-world coordinates. This holistic approach to surveying and modeling aims to produce a detailed digital twin of the structure, offering a dynamic and interactive representation that captures the evolving state of the observed lesion over time. 5. Conclusions The integration of the automated data acquisition system presented by the authors represents a significant advancement in data analysis research for cost estimation purposes. This system enhances operational efficiency and data quality by enabling more detailed detection capabilities, thereby reducing inspection costs and times. In the specific case examined, the use of the automated system for structural degradation detection facilitates data collection and processing, thereby enhancing estimation accuracy and enabling continuous real-time monitoring. The study illustrated in the article presents potential of considerable interest in order to improve the reliability of estimates in the feasibility project phase. In fact, exploring correlations between degradation levels and cost variations