This research explores the efficacy of the YOLOR (You Only Learn One Representation) algorithm integrated with the Deep Sort algorithm for real-time vehicle detection, classification, and counting in Morelia, Mexico. The study aims to enhance traffic monitoring and management by leveraging advanced deep learning techniques. The methodology involves deploying the YOLOR model at six key monitoring stations, with varying confidence levels and pre-trained weights, to evaluate its performance across diverse traffic conditions. The results demonstrate that the model is effective compared to other approaches in classifying multiple vehicle types. The combination of YOLOR and Deep Sort proves effective in tracking vehicles and distinguishing between different types, providing valuable data for optimizing traffic flow and infrastructure planning. This innovative approach offers a scalable and precise solution for intelligent traffic management, setting new methodologies for urban traffic monitoring systems.

The Heritage Cracks dataset contains a total of 1,716 high-resolution images documenting superficial cracking in ignimbrite masonry and lime-based plaster surfaces from heritage buildings in the historic center of Morelia, Mexico, a UNESCO World Heritage Site. A total of 686 images were acquired under real field conditions using smartphone cameras along representative streets and architectural elements with visible deterioration, while 1,030 images were generated through data augmentation. The dataset was designed to support artificial intelligence applications for cultural heritage inspection, particularly object detection and computer vision-based damage assessment. All images were organized, preprocessed, automatically oriented, resized to 640 × 640 pixels, and manually annotated with bounding boxes by specialists in construction pathology. Crack selection considered visible damage associated with weathering, material degradation, structural adjustment, environmental exposure, erosion, overload effects, and biological activity. The dataset is divided into training, validation, and testing subsets; includes image, annotation files, configuration metadata, and documentation to facilitate reproducibility. Its structure allows direct use in deep learning workflows, including YOLO-based models and other object-detection architectures. By providing field-acquired images from real heritage construction systems, this dataset contributes to the development, evaluation, and benchmarking of non-invasive tools for crack detection, preventive conservation, and structural health monitoring. It is useful for researchers, engineers, conservation specialists, and developers working on AI-assisted inspection of vulnerable historic infrastructure in Latin American urban heritage contexts.