Reliable and detailed underwater scene sensing, and analysis has a fundamental role in numerous underwater activities drawing both scientific and economic interest. Computer vision and remote sensing have evolved impressively in the last decade, propelled also by advancements in deep learning, enabling unprecedented levels of robust and automatic information extraction from visual data. At the same time, there is increasing interest in visual based underwater sensing, however, deep learning methods are less explored in this domain. On the other hand, sensing modalities based on Single Photon Cameras (SPCs) and transient imaging are gradually maturing, having certain characteristics that allow sensing in challenging visibility conditions, as the ones of underwater environments. iSEAu aims to significantly advance the state-of-the-art of underwater scene sensing by bridging the gap in the use of data-driven methods in underwater perception, and by combining the respective advantages of SPCs, multispectral and conventional cameras. Investing on intensive knowledge transfer, the goal is to bring together the fields of computer vision, machine learning and remote sensing for optimally addressing the underwater visual sensing challenges. The project objectives address these challenges in two levels. The first concerns the development of methods for “removing the water” from underwater images by harnessing the power of learning-based methods, and the development of methods based on SPC transient imaging for perception in challenging visibility conditions. The second level concerns the adaptation and enhancement to the underwater domain of state-of-the-art methods for image-based extraction of structural and semantic information, and their field-testing considering representative application scenarios. In summary, iSEAu will provide novel data-driven methodologies and technological solutions to researchers, scientists and users for underwater sensing of unmatched fidelity.

The problem of tunnel fires is one of the most complex areas of fire research. Under fire, concrete linings spall resulting in the collapse of the tunnel structure causing a significant scourge in the economy, society and the environment. In summary, the design of tunnel concrete linings is based on thermal calculations, which ignore spalling. Based on such calculations new types of more durable, strong and hence denser concrete have been introduced on the market recently that are much more probable to spall due to their lower permeability. To increase the permeability of the concrete and ultimately its fire resistance, it is commonly suggested to add polypropylene fibres. However, these tend to decrease the strength of the concrete and potentially its durability. In FiRe2C I propose a new type of Fire Resistant, Fibre Reinforced Concrete to improve on the tunnel lining's performance and ultimately the post-fire structural stability of the tunnel. To enhance our understanding of the performance of the Concrete on a multi-scale level, I will employ a holistic approach between state-of-the-art numerical (Discrete Element Method) and full-field experimental methods (advanced material and fire testing & x-ray computed tomography), pushing the existing boundaries of our scientific knowledge. Specifically, I will study experimentally the effect of size, distribution and orientation of the fibres on the strength and fire-resistance of concrete linings, by employing full-field imaging techniques pre- and post-fire, which has not been done before. And finally, I will create a novel DEM model to predict the thermo-mechanical response of fibre-reinforced concrete, utilising for the first time quantitative 3D experimental measurements at different length-scales to validate accurately the material response.