One of the most innovative experiments on the planet is taking place in the rainforest of the Republic of the Congo. There, the sounds of animals are studied with acoustic sensors. And among all the sounds that are captured in the Nouabalé-Ndoki National Park and its bordering forest areas, the Elephant Listening Project initiative, in which Microsoft actively collaborates with its Azure cloud platform, focuses on studying the bars of the African jungle elephants.
Why? African bush elephants are in serious danger of extinction. The scientists responsible for the Elephant Listening Project estimate that the African elephant population has decreased from around 100,000 animals in 2011 to less than 40,000 today, due to poaching for ivory for their tusks. In the last seven years alone, the elephant population in the Congo has decreased by 30%, mainly due to hunting, according to results published in 2016.
However, a project promoted by Microsoft within its AI for Earth project uses artificial intelligence (AI) and machine learning (ML) to monitor wildlife and evaluate the results of conservation work. With the Conservation Meetrics initiative, scientists and specialists are working to save this species, focusing on maintaining the balance of the ecosystem and attracting sustainable tourism.
Conservation Meetrics uses advanced algorithms to develop the Elephant Listening Project in collaboration with Cornell University, which is responsible for distinguishing the African jungle elephant bumps from the rest of the typical sounds of a tropical forest. It is a job that would take humans years to search for patterns between terabytes of data. Researchers use the data from the elephants’ calls to estimate their population, monitor their movements, try to ensure their safety and identify them individually. Analyzes include using ML to detect movement patterns in real time that alert to the arrival of poachers, or to block online advertisements about the ivory trade.
Conservation Meetrics began collaborating with the Elephant Listening Project in 2017 to boost its efficiency, and its AI-based algorithms and machine learning have been able to accurately identify elephant bumps. However, the volume of data they handle was collapsing the organization’s servers. For this reason, Microsoft AI for Earth will subsidize Conservation Meetrics for the next two years to develop a working method based on Microsoft’s cloud platform, Azure, with the aim of analyzing, processing and drastically accelerating the time in which they are analyzed. all data. Currently, computers take approximately three weeks to process several months of sounds. As soon as the migration is complete, by the end of the year, this process will take only one day.
Microsoft is also collaborating with the Peace Parks Foundation, which fights against the poaching of rhinos and other animals in South Africa, to create a remote sensing system that analyzes the risks of hunters and sounds the alarm. And he is helping to migrate to his cloud a tool that is used in dozens of conservation ecosystems across Africa to improve the effectiveness of patrols that protect wildlife.
Through the AI for Earth initiative, Microsoft has also collaborated with researchers from the USC Center for AI in Society (CAIS) and Carnegie Mellon University, who have created Protection Assistant for Wildife Security (PAWS). With this solution, they use artificial intelligence and machine learning to create route patterns for patrols, indicating the places where poaching is expected. CAIS has also created Systematic Poacher Detector, which detects hunters and animals in images through night drones and can also predict and find solutions in near real time to save animals.
