The Ministry of Public Works (MOP) is responsible for preserving Chilean heritage buildings, using a documentation process that gathers general information about these structures. This process is often lengthy and typically occurs when the property has already deteriorated. Currently, damages are represented using sketches and drawings, which do not accurately reflect the building's condition. We propose a new method for damage detection using Unmanned Aerial Vehicles (UAVs) to capture images of heritage buildings and Machine Learning algorithms for damage analysis. Our methodology begins with UAVs capturing images, which are used to build an orthomosaic of the building’s façade. This orthomosaic is divided into smaller sub-images that are processed by a pre-trained algorithm to identify known damage types. The images with detected damages are then reassembled into a final orthomosaic showing damage information. This method focuses specifically on damage detection, significantly reducing documentation time and improving the accuracy of damage representation. Our case study, Templo Votivo de Maipú, a 96-m-high reinforced concrete heritage building, was chosen due to its prevalent humidity stains. We validated our methodology through a survey of stakeholders, including the Templo management team and professionals from the Architecture Department at MOP. The survey, completed by nine respondents, rated the research an average of 3.9 out of 4 on a Likert scale (1: strongly disagree, 4: strongly agree). The final proposal was refined based on the feedback and recommendations from the validation survey.