The aim of this project is to explore new technologies that can improve the state-of-the-practice in structural health monitoring of existing bridges and infrastructures. In this context, vision-based approaches represent a powerful and cost-effective alternative to conventional devices. The present research proposal aims at developing and validating vision-based approaches for the structural health monitoring and assessing the structural safety of bridges in the context of digital transition in civil engineering. To this end, the project team includes complementary skills related to structural engineering, image processing, artificial intelligence, and signal analysis. In detail, image processing algorithms for cracks detection are developed by combining machine learning approaches, neural networks and photogrammetry to provide a semi-automatic workflow for detecting, identifying and measuring the crack patterns. Moreover, dynamic identification algorithms are developed through the processing of video, combining computer vision algorithms, motion magnification techniques and Data Fusion with accelerations acquired on the bridge. Data fusion techniques allow aggregating time series obtained by the video processing and those recorded by accelerometers placed on the structure to strongly increase the accuracy of the obtained displacements without significantly increasing the cost of the experimental equipment. Finally, an integrated methodology for the management of data and multi-level safety assessment of infrastructure is implemented. Based on the crack pattern recognition, the damage level of the structural elements is recognized and used for a preliminary qualitative assessment. The dynamic properties derived from vision-based identification are then used for refined assessment through the comparison with numerical simulations. The application to case-studies allows to verify the measurement accuracy, resolution, and robustness of the proposed approach, by addressing the sources of uncertainties and controlling the noise sources. A laboratory experimental case study will be developed and analyzed for a preliminary verification while a concrete road bridge in Modena will be chosen as a full-scale case study to validate the proposed procedures in the relevant environment. The goal of the project is to drive the applications of cost-effective sensors combined with AI-based strategies in the actual workflow of civil engineering, fostering the digital transition, introducing a strong innovation in the practice of safety assessment and tangibly contributing to promote a preventive approach in the management and maintenance planning of existing infrastructures.