Using AI to protect the Amazon

The Amazon rainforest in South America is one of the world’s largest and most dense, but it is disappearing at an alarming rate. Estimates from researchers indicate that approximately 18% of the Amazon rainforest has been completely lost, while an additional 17% has experienced degradation. Technology can help. Researchers are using AI to anticipate risk areas and implement actions to avoid deforestation before it’s too late, while decreasing its devastating effects on climate.?

Microsoft’s AI for Good Lab has been working for years to use technology to find solutions to our climate crisis such as the rapid deforestation of the Amazon rainforest, including in Colombia and Brazil. Spanning nine countries, the world’s largest rainforest covers roughly 2.7 million square miles and plays an important role in the regulation of global weather patterns.?

According to research by the nonprofit organization Amazon Conservation, almost 2.7million acres of the rainforest were subject to deforestation in 2022, a 21% increase from the year before, as land is cleared for farming or illegal mining operations. According to international environmental experts, if deforestation remains unchecked, it could permanently skew the planet’s ecosystem.?

This is an area of the world that we need to save, and solutions are best formulated when the scope of the problem is well-understood. Project Guacamaya is a collaboration by several organizations in Colombia in conjunction with our AI for Good Lab, including the public-sector Alexander von Humboldt Institute, the CinfonIA Research Center at Universidad de los Andes, and the Instituto Sinchi.??

This multimodal AI project monitors the Amazon in order to better understand the symbiotic relationship between forestation and wildlife.?

Satellites from technology partner Planet Labs PBC provide daily high-resolution images via the NICFI program, allowing Colombian authorities and scientific organizations to easily monitor the Amazon’s changes from space. One telltale sign that deforestation or mining is about to occur is the presence of unauthorized roads. If those are spotted in satellite imagery, authorities can be alerted to the activity. Project Guacamaya is developing AI models to analyze these images over time, spotlighting areas where illegal deforestation or illegal mining may take place.??

Next, Project Guacamaya incorporates data from tree-mounted cameras, known as camera traps, that have been placed throughout the Colombian rainforest. Anytime the cameras see movement, they take a picture. A single camera can take up to 300,000 photos.??

The Department of Biological Sciences at Universidad de Los Andes has contributed 110,000 images collected over the past four years. With the help of AI, researchers trained a model that detects which images contain animals and those that do not, and then classifies the animals in those images with 97% accuracy. Now, only 10% of images need to be manually validated, which not only allows researchers to expand and deepen their research but offers early indication of ecological shifts that need to be addressed.??

The final element of Project Guacamaya is sound. Using bioacoustics, researchers can capture sound from the Amazon and use an audio AI model to differentiate between birds and non-avian wildlife in a way that would be too labor-intensive and even costly for researchers to do manually.??

As the data and results from Project Guacamaya continue to grow and be refined, our hope is for other conservationists and countries will leverage these tools to help protect the Amazon throughout Central and South America.?

This is the second installment in our series on how AI can be a force for good and help solve the world’s most pressing problems.?