![](https://www-smartinfrastructure.eng.cam.ac.uk/sites/default/files/styles/leading/public/vlad-workshop-story-883x349_copy_0.jpg?itok=PdEPtyHU)
Submitted by L. Millard on Wed, 16/12/2020 - 14:35
In this online interactive workshop CSIC Researcher Vladimir Vilde will discuss the different potential challenges faced during the selection, installation and the processing of computer vision (CV) techniques.
Leading the workshop, which will take place on Thursday 4 February 2021 from 9.30am to midday, Vladimir, who is researching novel solutions for infrastructure monitoring using digital twin and computer vision techniques, will cover examples of CV techniques that can be used in construction by addressing limitations and opportunities.
Innovations using CV in infrastructure often remain at the case study stage. Issues such as scalability, funding for long term usage, or suitability for industrial application can limit widespread adoption of new and promising technologies.
Recent innovations in CV solutions have been very attractive for their accessibility and potential benefit. Deep Learning techniques are very popular in CV and can be found in applications from self-driving cars to cancer detection. Projects in construction have tried to emulate these successes, but despite returning interesting results early on, these AI techniques are known to be expensive when considering robust usage.
Currently, there are too many projects in construction trying to rely solely on AI, while conventional computer vision is not properly considered. Each challenge can have a unique answer – it is unlikely that one solution fits all. This is especially true in construction and asset management, where funding can be limited: the technologies not only need to be smart with the data, but with the budget as well!
Imaging techniques can be complex and are widely applied, from smartphone cameras to satellites, so considering their use on a project can appear overwhelming. They are often used as “black boxes”, where AI is expected to work exactly as intended. However, smarter and more adapted solutions might be readily available if we have the tools to open these “boxes”.
Who should attend?
Managers involved in infrastructure sensor deployment decisions with an interest in deploying computer vision-based solutions.
Registration: Registration is now open for the free ‘Computer Vision Beyond Black Boxes’ workshop. Please RSVP to csic-comms@eng.cam.ac.uk if you would like to attend.
Date: Thursday, 4 February 2021 from 09:30 to 12:00