ISEL - Matemática Aplicada para a Indústria
Permanent URI for this community
Browse
Browsing ISEL - Matemática Aplicada para a Indústria by Subject "Artificial intelligence"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
- The use of artificial intelligence in the recognition of railway assets based on high-resolution drone imagesPublication . Robu, Cristian; Geraldes, Carlos José Brás; Cal, Filipe SantiagoAbstract Ensuring the safety and operational continuity of railway infrastructure requires precision, efficiency, and reliability in maintenance and management practices. Traditional methods of asset management, characterized by manual processes and high error rates, often fall short of these requirements due to their labor-intensive nature and significant costs. This thesis aims to overcome these challenges by integrating advanced artificial intelligence (AI) technologies into the asset management process, with a particular focus on deep learning for automated object detection. This research explores the application of YOLOv9, the latest iteration of the You Only Look Once object detection algorithm, for identifying and cataloguing railway assets from high-resolution drone-captured images. Drones provide a unique aerial perspective, enabling comprehensive monitoring of extensive and often inaccessible railway areas. YOLOv9 is especially well-suited for this task due to its efficiency in processing large volumes of high-resolution images and its robustness in delivering real-time, high-accuracy detections. The study involved training the YOLOv9 model on a dataset of drone-captured images, which were carefully annotated with the locations and identities of various railway assets. The performance of the model was assessed by measuring its precision and recall in detecting these assets under a wide range of different environmental conditions. The findings reveal that AI can improve the precision and efficiency of railway asset management. The YOLOv9 model significantly reduced the time required for asset inspections, thereby contributing to safer railway operations. This approach highlights the scalability and flexibility of AI technologies in infrastructure management, offering valuable insights into their potential for broader applications across various sectors. The findings underscore the transformative potential of AI in improving the management of critical infrastructure and lay the groundwork for future research in this rapidly evolving field.