Smart CCTV systems that are designed to identify blockages in city waterways may play a crucial role in preventing floods in the future, according to new research.
Camera systems powered by machine learning have shown to be an efficient and cost-effective method for flood defense, as demonstrated by researchers.
Recent findings from researchers at the University of Bath indicate that smart CCTV systems can be programmed to recognize natural debris, litter, or waste that obstructs trash screens in culverts, which are key in flood prevention efforts.
Culverts, which exist in over a million locations across the UK and are essential in almost all urban environments, enable streams and rivers to flow beneath roads, railway embankments, and buildings. This makes them an integral, albeit often overlooked, component of urban waterways and infrastructure. Trash screens, typically consisting of bars, are positioned at the openings of culverts to prevent debris from passing through.
Blocking a trash screen at a culvert can lead to quick flooding. If the water entering a culvert is obstructed, it can accumulate and pool, potentially compromising the waterway’s structural integrity and impacting the surrounding environment.
Enhancing global flood defenses
The machine learning model developed by the research team is garnering attention from flood management organizations in various countries, including South Africa, which have monitoring systems but lack the necessary data for training AI to perform similar tasks.
Dr. Andrew Barnes, a computer science lecturer at Bath and a member of the Centre for Climate Adaptation & Environment Research, highlighted, “We have engineered an effective model that identifies blockages before they escalate into a problem—it’s proactive and alerts authorities before flooding occurs.”
“Our system is designed to be adaptable and scalable, making it applicable in various settings. This flexibility allows it to be beneficial in countries where flooding poses a risk but resources to create local solutions might be limited.”
Accurately detecting obstructions
By focusing on a culvert in Cardiff, the team trained a camera system using machine learning to autonomously detect possible blockages, achieving nearly 90% accuracy in identifying these hazards. In the UK, most culvert monitoring is currently done manually, with local authorities observing multiple CCTV feeds.
Implementing AI and machine learning for early warning systems would enable local authorities responsible for maintaining clear waterways to allocate resources more effectively and respond to blockages swiftly and precisely.
The proactive design of the system also enhances safety for response teams, allowing them to address issues immediately, rather than under dangerous, flooded circumstances.
Dr. Thomas Kjeldsen, a Reader in Bath’s Department of Architecture & Civil Engineering and member of the Centre for Regenerative Design and Engineering for a Net Positive World (RENEW), noted, “With climate change increasing flooding risks globally, this work paves the way for innovative, lightweight, and cost-effective flood management systems in urban areas, enabling authorities to better adapt to climate challenges. This research marks an initial move towards sustainable flood forecasting solutions and opens numerous avenues for further study.”
