Member-only story
Detecting Polluting Industries with Machine Learning
Satellite imagery analysis and neural networks could offer a viable and efficient solution, as shown by researchers from the University of Lahore and Oxford.
As a Data person, I feel very comfortable with the idea that solving the Carbon emissions problem is first and foremost an engineering task. A huge one, but still something that must be approached that way. Solving a complex problem is usually about splitting it into smaller, more manageable sub-problems, making measurements, and planning for solutions.
Industries contribute about 25% of the total Carbon Dioxide emissions according to the International Energy Agency and the biggest challenge is tracking those down to reliably estimate individual impacts. This doesn’t just mean the big industrial conglomerates, which are easier to identify and monitor, but also of the myriads of small-scale, often unregulated industries that are scattered in rural areas of poorer countries.
Computer Vision and Machine Learning can help with scanning thousands of satellite images to find the exact location of those industries, as demonstrated in a recent paper by the University of Lahore and Oxford. Researchers applied a promising methodology to the…
