More about this talk follows:
Condition assessment of civil infrastructure is a key instrument for infrastructure managers to evaluate structural integrity and operability, as well as defining possible maintenance or rehabilitation strategies. In recent years, remote inspection techniques based on computer vision and Unmanned Aerial Systems (UAS), also known as drones, have been recognized as key components for improving inspection and monitoring strategies to achieve an automated condition assessment of civil infrastructures. These technologies proved to be competitive in identifying damage in inaccessible and extensive areas, allowing a considerable reduction of costs and execution times. This presentation is focused on the latest developments on the remote inspection of civil infrastructures using advanced image processing techniques based on Artificial Intelligence. Within this topic, Deep Learning algorithms, such as the Convolutional Neural Networks (CNNs), and its latest enhancements, like the Mask R-CNN algorithm, will be detailed. The application of these AI algorithms to the automatic damage identification on large scale infrastructures will be presented. The first case-study is focused on the detection of exposed steel rebars in a storage silo, while the second case-study is related to the detection of corrosion on roofing systems of industrial buildings.