AI against Toxic Clouds

Use Computer Vision to detect toxic clouds emitted by large factories and report the authorities on the sightings


Premium participation: Get 1-on-1 Mentoring & Guidance

Try a free mentoring session

Help environmental authorities monitor illegal air pollution by large factories.

Illegal pollution damages the environment and public health. Old production plants often do not reach the standards of today's laws and regulations, but, due to corporate lobbying and meager funding, authorities are rarely able to uphold these standards. To combat toxic emissions, local governments can issue fines for harmful incidents. The only problem is that some companies cross the line daily, while environmental agencies do not have the manpower to monitor a facility 24/7.

In this Challenge, we will collaborate with Greenpeace to aid in the monitoring of the Tata Steel plant in IJmuiden,  the Netherlands.

Tata Steel* is one of those sites that regularly release illegal toxic substances into the air. For this specific steel plant, we narrow down the problem to the most polluting factories, which are the two coke ovens*. These are extremely old and leak like a sieve.  When processing raw coke, the factories sometimes release black clouds containing carcinogenic PAHs (polycyclic aromatic hydrocarbons), which is illegal under Dutch law. Unsurprisingly, the area around the Tata Steel plant sadly has a significantly higher incidence of lung cancer than the rest of the Netherlands.

Tata Steel is obligated to report incidents in which PAHs escape to the Dutch Environmental Services but rarely does so. The Environmental Services do not have sufficient capacity to register and follow up on all incidents. Tata Steel seems to disregard the fines they receive when the Environmental Services confirm the release of PAHs. They simply pay the fine of €100,000 and continue to emit harmful substances. So far, they have received three fines, but when they reach fifteen fines, their permit will be revoked, and the Coke ovens must be shut down.

Residents of IJmuiden, organized under the foundation FrisseWind.nu, have installed cameras to monitor the coke ovens and other parts of the Tata Steel plant. They report any suspicious activity to Environmental Services when they observe something unusual. However, they cannot watch the footage continuously. The Environmental Services themselves have also aimed a camera at one of the coke ovens, but it is unlikely the staff is able to monitor the footage all day long. By enlisting the help of the public, via the Spot De Gifwolk campaign, Greenpeace makes sure there are enough people to collectively watch the webcam streams At the same time, this allows them to compile a large, labeled dataset of illegal emissions.

While volunteers from Greenpeace continuously look for violations we will be gathering a team of AI engineers that aim to monitor the video streams with computer vision models. In this way, the efforts of the people spotting toxic clouds will be immortalized. The more violations we report, the sooner the dangerous Coke ovens will have to close. 

* Tata Steel Nederland is the Dutch branch of the Indian steel producer Tata Steel. The steel mill has been located at the current site in IJmuiden since 1918 and this facility is the third-largest mill in Europe and the fifth most polluting.

* Coke is a hard, and porous coal-based fuel, made by heating coal or oil in the absence of air, used mainly in iron ore smelting.

Info session

Our challenge partner

GOAL: Detect toxic clouds emitted by large factories on a continuous video stream and report their sightings

By reporting illegal pollution to Environmental Services. This Challenge might lead to more fines and, ultimately, the revocation of their environmental permit, resulting in the closure of the coke ovens. 

Additionally, the resulting model could be applied more broadly to monitor polluting factories all over the world.  

In this Challenge, we will 

  • Use video data to create a dataset large enough to train CV models
  • Build classification models that detect illegal pollution
  • Deploy the CV models on a live video stream  

Technologies we will use

  • Data wrangling
  • Image classification
  • Transfer learning
  • Streaming data pipelines 

Who are we looking for? 

We are looking for data science & AI engineers, previous work with computer vision and model deployment would be of great benefit.

You will collaborate with a diverse team of up to 50 international collaborators in subteams. You can join as a contributor (8-12 hours per week commitment for 10 weeks) or coach (2-4 hours per week, only for experienced ML professionals)

We’ll organize a masterclass on smoke detection and video analysis with AI during the challenge to bring you up to speed.   

Did you know

🇳🇱 Lung cancer is up to 51 percent more common in the immediate vicinity of Tata Steel in IJmuiden than in the rest of the Netherlands.

🏭 The steel factory in IJmuiden started producing steel in 1920. The factory was a proud establishment from the Koninklijke Nederlandse Hoogovens en Staalfabrieken (Royal Dutch Blast Furnaces and Steel Factories). In 1924 the first Coke oven was taken into use. In 1926 the second followed.

Application deadline

July 17, 2023
To application page


Application Deadline: 17 July 2023

Challenge Kick-off: 19 July 2023

Midterm Presentations: 9 Augustus 2023

Final Presentations: 27 September 2023

A little more