Congratulations to Maurice Fallon and the team at Oxford University on the award of this project. This project focuses on the development, testing and demonstration of omni-directional cameras (“360 cameras”) for long-term visual facility monitoring. These high-quality cameras have been developed for use by online content creators which have driven the quality of sensors while retaining low cost. However the data output is raw, unprocessed image footage. A variety of visual mapping technologies are required to organise and collate recordings to provide end-user value – from well-studied mapping techniques to emerging scene understanding techniques. In Omnivision Connect the team will develop multisession visual mapping (SLAM) and visual AI algorithms to organise and collate the data recordings to give end-users insights about their facilities – features such as detection of objects of interest and their movement across time – to enable progress monitoring on large facilities such as Sellafield.
During the project the team will collect a long term dataset of recordings in an industrial facility similar to those in the nuclear sector – with poor lighting, fast motion and regular changes. Insights will be extracted using Visual AI and present to end users via a browser-based UI.
