PLOTO’s WP5 aims to develop approaches that early detect damage and degradation and satisfy the diverse information requirements while make use of state of the art and emerging computer vision and ML techniques. IWW elements can be divided into open, semi-enclosed and enclosed sections, such as tunnels, avalanche galleries etc. Apart from the IWW elements, the ports which are the most critical infrastructures along the corridors, are exposed to extreme weather conditions among other issues. In fact, IWW and ports’ elements and infrastructures are in great need of a pilot monitoring system on which CV and ML techniques will be applied, to provide damage diagnosis for key infrastructures. In the lifetime of PLOTO project deep learning, machine learning methodologies and tensor Algebra decomposition will be enhanced in order to perform damage diagnosis and crack detection on the existing infrastructures and IWW elements along with information fusion techniques using different data modalities i.e. satellite images, multispectral data, LiDAR, RGB images etc. New methodologies will be developed to identify, characterize, quantify and monitor damage. In particular, the existing CNN-based classifiers will be initialized with computer vision features focusing on the precision in recognition, while the crack detection algorithms will be refined by augmenting the CNN with 3D information (point clouds, 3D meshes) and restrictions for cracks. In addition, the 3D data will assist on the extraction of structural defects, i.e. distance between cracks, length etc. The detection of structural deformations and other types of alterations in 4D manner, and outlier and anomaly detection techniques will provide valuable information for the infrastructure’s health in both everyday life and of course post-disaster scenarios.
Written by Margarita Skamantzari, Researcher, Lab. of Photogrammetry, National Technical University of Athens.