Visual pollution: what it is and how AI detects it

Visual pollution is the man-made clutter and decay that make a street look neglected: unauthorised billboards, graffiti, construction waste, potholes, broken pavements, abandoned cars. It is hard to manage because it is everywhere, it changes every day, and inspectors only see the streets they reach that week. AI changes that by turning the cameras a city already has into a continuous inspection: every image checked, and every problem classified, located and sent to the team that fixes it.

What counts as visual pollution

The term usually describes the negative effects of man-made structures on how people perceive and enjoy their surroundings. In practice, cities treat it as a list of specific, fixable problems:

• Unauthorised billboards, signs and advertising
• Graffiti and defacing writing on walls and public property
• Construction and demolition waste, and excavations left open after roadworks
• Potholes, damaged roads and broken or blocked pavements
• Abandoned and damaged vehicles
• Overflowing bins and illegally dumped waste
• Broken street lights, exposed wires and open manholes
• Dilapidated facades and unauthorised structures such as sheds and canopies

Some of these are only eyesores. Others are hazards: an open manhole or an exposed cable is a visual problem and a safety problem at the same time. That overlap is why visual pollution usually sits with municipal and public safety teams, not only with urban planners.

In Saudi Arabia the same idea is called visual distortion (التشوه البصري). The national Visual Distortion Treatment Initiative, launched in 2018, sets out to remove 17 manifestations of it from the urban landscape.

Why visual pollution matters

It is easy to file visual pollution under aesthetics. The research doesn’t. Long-term exposure to unpleasant or intimidating streetscapes is associated with stress, people walking less, distracted drivers, lower satisfaction with where people live and, in some cases, lower property values.

For a city the cost is concrete: clean-up budgets, complaints, injuries on broken footpaths, and violations that are never recorded and so never enforced.

Why cities struggle to keep track of it

The problem is scale. In 2021 alone, Saudi Arabia’s municipal ministry and its municipalities removed:

• 120.9 million cubic metres of construction and demolition waste and excavation spoil from vacant land
• 1,066,463 unlawful advertising signs
• 116,638 damaged vehicles
• Defacing writing covering 2.8 million square metres

In the same year they repaired 15.9 million square metres of potholes, 4 million metres of broken pavement and 876,018 street light poles.

Every one of those had to be found before it could be fixed. Before AI, inspectors in the Saudi programme physically covered around 10% of public space. Citizen reports help, but they only arrive once a problem is bad enough to report. At any given moment, most of what is out there is unrecorded.

How AI detects visual pollution

Visual pollution detection with AI works as a pipeline, from camera to work order:

1. Capture from cameras the city already has. CCTV, dashcams on municipal and service vehicles, inspector and citizen mobile apps, and drones. Starting doesn’t require new hardware.
2. Detect and classify. Computer vision models find each problem in an image and label it by category, subcategory and severity. On CamCom’s platform this takes under 60 seconds per image.
3. Locate it and add context. Every detection is geo-tagged and checked against GIS rules: whose jurisdiction it falls in, what the land is used for, what else is nearby. A pothole outside a school is a different priority from the same pothole on an industrial service road.
4. Send it to the right department. Each incident is matched to the relevant municipal regulation and routed to the team responsible, with the photo, location and time as evidence.
5. Track it until it is fixed. Maps and history show where problems repeat, so maintenance budgets go where the problems actually are.

The hard part is not spotting a pothole in a clean photo. It is staying accurate across lighting, weather, camera angles and cluttered scenes, at the volume of a whole country, and covering dozens of categories in one system rather than one model per problem. Public research reflects the same priorities: a 2023 benchmark built from 34,460 images collected through the Saudi ministry is labelled for three classes: excavation barriers, potholes and dilapidated sidewalks.

Visual pollution AI at national scale: Saudi Arabia

Saudi Arabia treats visual pollution as national infrastructure rather than a local clean-up job. The work is led by the Ministry of Municipalities and Housing, known until July 2024 as the Ministry of Municipal and Rural Affairs and Housing (MoMRAH).

Under the Quality of Life programme within Saudi Vision 2030, the ministry deployed CamCom’s AI platform to detect and classify public safety and visual pollution elements in images from citizens’ mobile reports and field surveys. Each incident is classified, mapped to the relevant regulation and assigned to the responsible department, with alerts to the officers concerned and a national view in a central command centre. The partnership was announced in June 2023 as a global-first programme to tackle visual pollution with AI.

Results CamCom has published from the deployment:

• 30 million+ incidents processed across 16 municipalities
• 450,000 km of road documented, every 50 metres
• 43+ categories detected, up from 6 at launch
• Case processing cut from three weeks to under one week
• A 62% reduction in fatalities and casualties

Read the full Saudi Arabia case study.

Five questions to ask before you buy visual pollution detection

1. Does the category list match your regulations? Every city defines visual pollution in its own codes. The system should map each detection to your clauses and add new categories without a rebuild.
2. Does it work with the cameras you already have? If a pilot needs new hardware, scaling it to the whole city costs more and takes longer.
3. How is accuracy measured? Ask for figures from live deployments, broken down by category and severity, not one overall score on a test set.
4. Who checks a detection before it becomes a fine? AI can assemble the evidence. A person in the responsible department should confirm the violation, and that review should be recorded.
5. Where does the data live? Government imagery often has to stay in the country. Check for on-premises or in-region deployment and compliance with local law, such as Saudi Arabia’s PDPL or India’s DPDP Act.

For cities and ministries, CamCom offers this as Vigil by CamCom: AI hazard detection for cities, covering 43+ categories of urban hazard from the cameras a city already has.

Common questions

What is visual pollution?
The man-made clutter and decay that degrade how a place looks and feels: unauthorised billboards, graffiti, construction waste, potholes, broken pavements, abandoned vehicles and overflowing bins.

What are examples of visual pollution?
Illegal signs and billboards, graffiti, construction and demolition waste, abandoned cars, dumped rubbish, damaged roads and pavements, broken street lights and dilapidated facades.

What is visual distortion in Saudi Arabia?
Visual distortion (التشوه البصري) is the term Saudi Arabia uses for visual pollution. The Ministry of Municipalities and Housing’s Visual Distortion Treatment Initiative, launched in 2018, targets 17 manifestations of it and gives citizens a way to report cases to the authorities.

How does AI detect visual pollution?
Computer vision models analyse images from CCTV, vehicle dashcams, mobile apps and drones. Each problem is classified by category and severity, geo-tagged and routed to the department responsible.

Can AI detection be used to issue fines?
AI can produce the evidence for an enforcement case: the photo, location, time and category. The decision to issue a penalty should stay with the responsible department, with a person reviewing the evidence.

Sources: Paulo Anciaes, “Visual Pollution“, in M. Garrett (ed.), Encyclopedia of Transportation: Social Science and Policy, SAGE, 2014. Ministry of Municipalities and Housing, 2021 visual distortion figures (Arabic), 15 May 2022. Saudipedia, Visual Distortion Treatment Initiative. Saudi Press Agency, ministry renamed, 21 July 2024. Data in Brief, Visual pollution real images benchmark dataset on the public roads, October 2023. CamCom Saudi Arabia case study.