Stanford Researchers are employing geospatial artificial intelligence technology that combines satellite photos with algorithms that can detect global regions where poverty is most acute.
Experts previously have employed nighttime satellite images to identify impoverished areas. When populated areas lack electric lighting, it can serve as a telltale sign of destitution.
For example, a famous satellite image of the Korean Peninsula shows the dramatic contrast between the bright and prosperous South Korea and the dark and deprived North Korea.
However, this indicator can’t measure the level of poverty to help find the areas that are most in need of aid.
Enter Stanford University economist Marshall Burke who is developing a system that can identify features indicative of the greatest level of hardship. As reported by NBC News, Burke and his colleagues are also using daytime satellite images and applying AI techniques that can provide information on poverty.
“Instead of us hand-curating the data and telling the computer what features to look for, we wanted to allow the computer to figure it out on its own,” Burke said, in a quote in the NBC article.
The Stanford team gave the algorithm both nighttime and daytime imagery from the African nations of Uganda, Tanzania, Nigeria, Malawi, and Rwanda, which have household survey data. The researchers then asked the algorithm to identify features in the daytime imagery that correlate with places that are illuminated at night.
The algorithm noted specific features that are associated with nighttime lighting and identified patterns that could distinguish poverty. Based on the survey data, the algorithm was able to predict poverty in an area with 81 percent to 99 percent better accuracy compared to using only nighttime images.
Based on this information, government agencies could monitor the economic conditions in poor regions, as well as find places that need the most aid, NBC notes.
Satellite-based geospatial AI also is being employed to fight poverty in other ways. One approach is to use the technology to predict crop production, possibly anticipating and mitigating the impact of food shortages.
Boston-based startup TellusLabs is using a mix of satellite imagery, weather data, artificial intelligence and human expertise to predict crop production. This technique allows the company to observe and predict the impact of events like drought or natural disaster on crops. The company said it has been able to consistently predict the USDA’s final 2016 corn and soy yield reports in advance of public in-season forecasts.
Stephan Cunningham is vice president, product management at Veritone. Working in concert with core internal teams including industry-specific general managers and engineering as well as directly with clients and prospects, he leads the disciplines and business processes which govern the Veritone Platform.