Satellite Imagery, Crowdsourcing and Machine Learning
We live in an age where artificial intelligence is increasingly being deployed to interpret satellite imaging. But the methods deployed are still in their infancy and – very costly. What if we could develop a system that acts as a catalyst to enhance these methods that at the same time helps to tackle some of our great societal challenges?
Enter Crowds and Machines, a new project that uses satellite information in a unique way to support the development of more effective interventions to mitigate the negative impacts of events that can be observed and monitored from space.
Our starting point: Cerberus. Cerberus is an advanced gaming platform that uses crowdsourcing to interpret satellite imaging. It currently utilises the eyes of over 75,000 people to build advanced mapping products. The output: detailed maps that are better than Google.
How? Using trained players, the generated data is used to train AI and machine learning systems that can detect pre-identified objects and scale the original maps to much larger surface areas.
Crowds & Machines shows us how collaboration between man and machine can be used for good.
Using the crowdsourcing tools of Cerberus and the passion for gaming of people around the world together with the latest machine learning techniques, we mapped, monitored, and measured the rate of production of cereals and other crops in Ethiopia and Italy in the period 2016-2021.
Our aim: to observe whether the covid-19 pandemic impacted yields and harvesting potential under varying climate conditions and enhance our understanding of the causal pathways between food security and political stability.
Ultimately, the method can be used to complement ongoing field work that is often costly, time-consuming and reliant on fragmented information. Moreover, since the data can be collected, interpreted, and scaled remotely, it can be applied to regions that are far removed, difficult to reach, or dangerous to enter.