Natural hazards such as earthquakes, floods, and hurricanes pose severe risks to the structural integrity of concrete infrastructure, making rapid and reliable post-event damage assessment essential. Conventional inspection procedures rely primarily on manual, on-site visual evaluations conducted by structural engineers; however, these approaches are often time-consuming, labour-intensive, and potentially unsafe when dealing with unstable or partially collapsed structures. Recent advances in unmanned aerial vehicles (UAVs) and artificial intelligence (AI) offer a promising alternative to traditional assessment methods. UAVs equipped with high-resolution imaging sensors enable rapid and safe data acquisition in hazardous or inaccessible environments. When coupled with AI-based image analysis techniques, these systems can autonomously detect and classify structural damage—such as cracking, spalling, and delamination—and support a quantitative assessment of damage severity. This study proposes the development of an AI-driven UAV-based inspection framework for post-hazard evaluation of concrete structural elements. The framework integrates UAV-based data acquisition with advanced deep learning techniques, including convolutional neural networks for damage detection and classification, and data-driven surrogate models for residual structural strength estimation. The neural networks are trained on large datasets generated from prior Finite Element simulations of fractured structural components following a confined concrete damage plasticity algorithm [1]. Damage identification is performed through a voxel-based, piecewise segmentation of the structural component into cubic sub-domains of 10 × 10 × 10 cm. For each sub-domain, damage classification is carried out and the corresponding strength reduction is estimated. A phenomenological law for residual strength quantification, derived from single-block response data, is formulated and compared with the predictions of the trained network, showing satisfactory agreement. The proposed framework demonstrates strong potential as an automated, efficient, and reliable decision-support tool for post-disaster structural assessment, enhancing inspection safety and contributing to improved infrastructure resilience. REFERENCES [1] G. Mazzucco, B.F. Dongmo, B. Pomaro, J. Zhang, V. Salomoni, C. Majorana, A 3D visco-elasto-plastic damage constitutive model of concrete under long-term loads. European Journal of Mechanics - A/Sol