Building maintenance is often seen as a low-priority financial burden, which is a paradox, since given that the built environment corresponds to approximately 50% of the wealth of most European countries. Buildings are one of the most valuable assets of any individual or collective entity, public or private. Reactive maintenance, based on subjective criteria, is still current practice. In a Society with scarce resources facing the challenges of economic and environmental sustainability, typified measures for the maintenance of buildings can no longer be applied. Currently, paper-based records and digital spreadsheets are still the best approach used in buildings’ management, resulting in inefficient operations and maintenance. Building information modelling (BIM) is a practical approach to store, visualize and exchange building information in design and construction, but it is still at its beginning concerning the operation and maintenance phases, which comprise around 80% of whole life-cycle costs. To overcome the current limitations, image-based methods for the inspection of buildings’ facades can be very effective, particularly when combined with advanced technologies such as computer vision and machine learning. The image-based methods allow a promptly capture of large amounts of data in a relatively short amount of time. This can lead to more efficient inspections, especially for a large set of buildings. The computer vision and machine learning algorithms can automate the analysis of images, making it easier to detect defects, anomalies, or critical areas. These algorithms can be trained on large datasets to improve accuracy and reliability over time. Factors such as lighting conditions, weather, and the presence of obstructions can affect the quality of images, and the effectiveness of analysis algorithms, being essential to take it into account. This paper presents a methodology to build a digital model of buildings to support management maintenance, by merging the damage maps by computer vision and a degradation index, both automatically obtained. The automatic mapping of facades is performed by a supervised classification, followed by the calculation of the degradation index, based on the maps obtained. The results are introduced in a digital model of the buildings (such as BIM) for a comprehensive analysis of the actual state of conservation on the building as well as its evolution during service life. The methodology is showcased with a real building, evaluating the impact of different maintenance strategies in the performance of the case study under analysis. The results demonstrate the applicability of the proposed approach to support decision making in terms of maintenance. The final digital model enables the integration of all data, being a valuable and useful tool for different stakeholders’, such as asset managers.
