“AI is Cracking Down on the Illegal Dumping of Rubbish”

The many problems of fly-tipping

Fly-tipping is a growing issue across the UK causing health and safety issues, a significant cost for local authorities and blighting communities. It provides a habitat for vermin and risks people’s health when hazardous waste is dumped such as asbestos, unknown chemicals, gas bottles and mattresses. Not only does this have a detrimental environmental impact but it encourages anti-social behaviour through the association with criminals, who deposit the material, and reducing the sense of pride in the area. All of which has a huge impact on nearby residents.

During lockdown, the illegal dumping of rubbish in the UK increased significantly. And the problem hasn’t gone away: it’s estimated that there were one million cases of fly-tipping in the UK last year. Social and operational teams at Kingston and Sutton’s councils have started an initiative using Artificial Intelligence (AI) technology to catch fly-tippers in the act.

A new Internet of Things (IoT) platform

Kingston and Sutton Councils collaborate with three other London Boroughs as part of the South London Partnership (SLP). Back in 2018, the SLP recognised that technology could be a powerful ally in helping with the problems faced by local authorities. They secured funding to develop an IoT programme which would focus on addressing real world problems using data and insights gained from IoT sensors deployed around various locations.
As the scale of the programme grew, the decision was made to expand the scope to include building a brand new IoT data platform. This would overlay data in ways not previously possible and give the potential for profound insights by breaking the data silos within the local authority community. It would also allow data to be more widely available to the public.

How can technology prevent fly-tipping?

Technology can act as a mechanism for both deterring and catching perpetrators in the act of fly tipping. Following an open tender on the SPARK CCS Framework, Vodafone and iDefigo were appointed. They proposed a camera solution that could determine when items had been deposited in monitored areas. From January 2022 this technology was deployed across known fly-tipping hotspots across Kingston and Sutton.
Each location requires a combination of cameras and wireless transmission bridges. The devices usually use the power supply of the lampposts that they are mounted on, or they can be powered by a solar panel which enables them to be placed on something without a power source, like a tree. The cameras don’t record continuously but provide a snapshot of several frames when they detect movement; this reduces energy consumption and limits the amount of footage needing to be reviewed.

The footage is wirelessly transmitted to the cloud based IoT platform where it’s analysed by AI. When the system believes there has been a case of fly-tipping, it sends an alert with supporting pictures to council enforcement teams. Despite the cameras being triggered thousands of times a day, the technology whittles down the footage to a very manageable number of highly likely fly-tipping cases.

The cameras capture images of the suspected perpetrator(s) together with the discarded items. This enables the enforcement teams to see if an offence has been caused and to arrange for a suitably briefed crew to be dispatched to remove the waste. The human images provide evidence to pass onto enforcement teams who can then issue a fine or take legal action if fly-tipping is confirmed. If not flagged for investigation, the data is deleted.

Is AI foolproof in detecting fly-tipping?

The vision technology used is extremely clever, but it can only reliably identify a particular object or shape like a human figure, or a box. Rubbish can be any number of items in varying shapes and sizes and so it does create a slight challenge.

iDefigo are continuously working behind the scenes to improve the technology’s ability to detect fly-tipping. Their data scientists review footage and check that the tagging the system gives to ‘rubbish’ is correct. This picks up instances where rubbish has falsely been identified such as when someone innocently walks past with a large bag or where it has failed to be noticed. This learning is fed back into the system to refine its accuracy in detecting fly-tipping and so it is continuously improved.

The complete picture can be accessed by operational teams at any time via a dashboard. After logging onto their device, they can view comprehensive footage and by skipping back or forward a few frames can more accurately assess the situation. The system integrates with smartphone apps that make it very easy for officers to report other cases of fly-tipping that are discovered when they are out and about. This can then provide evidence for new cameras to be installed somewhere to keep an eye on things.

What results have been seen so far?

The nature of fly-tipping is that it is often a small number of individuals who create the problem in a given location. So, by identifying even one offender using this technology, a significant proportion of the problem starts to be tackled. Some areas have been truly transformed by the introduction of these cameras. One estate has gone from having trucks regularly dumping huge volumes of commercial waste – to there being none. Another has calculated an impressive 80% reduction in fly-tipping.

Housing estates have seen fly-tipping levels notably reduce in the days after cameras and signage are installed. The devices simply being present acts as a deterrent, stopping it happening in the first place with potential fly-tippers spreading the word about these interventions. And then there is a snowball effect as by creating a cleaner environment, it breeds a newfound respect for the area. This starts to change behaviour, improving the location for the people that live and work there in multiple ways.

Find out more
If you would like to have a conversation about how this technology could help your vulnerable residents, please contact Andrew Parsons andrew.parsons@kingston.gov.uk or Pierre Venter pierre.venter@sutton.gov.uk.