Peru is home to some of the world's most biodiverse tropical forests. Yet illegal logging and the removal of endangered tree species is threatening its biodiversity and sustainable forest management.
The FORGE project, led by Jefersson A. dos Santos, aims to evaluate how artificial intelligence and drone-based remote sensing could improve the monitoring of tropical forests in Peru to help combat these threats. Supported by the Grantham Amplification Fund, the project brings together expertise in artificial intelligence, forestry, and conservation, to assess whether AI algorithms could accurately identify tree species using drone imagery validated with field data.
“Illegal harvesting not only threatens biodiversity but also undermines the enforcement of the Convention on International Trade in Endangered Species (CITES) and reduces the forests' ability to provide essential ecosystem services such as carbon storage, climate regulation, and water security,” explains Jefersson. “Rather than developing new AI methods, the research focused on evaluating the effectiveness of existing approaches in real-world forestry applications to support more accurate and scalable forest inventories.”
How were helping protect the Peruvian rainforest
The urgent conservation challenge within the Peruvian rainforest provides an unique opportunity for the team to make a real-world impact through a long-standing partnership with OSINFOR, the Peruvian agency responsible for supervising forest concessions.
“The project focused on detecting the Shihuahuaco tree (Dipteryx micrantha), a species listed under the CITES of Wild Fauna and Flora, to demonstrate how these methods could support biodiversity monitoring and the enforcement of sustainable logging regulations.”
Protecting these forests demands accurate and up-to-date forest inventories. However, traditional field surveys are expensive and time-consuming, requiring teams of specialists to work across vast and often inaccessible areas.
“Our collaboration with OSINFOR allows us to tackle these challenges together,” says Jefersson. “They provide field-validated forest inventory data, help identify priority study sites, and enable us to evaluate our artificial intelligence models under real operational conditions. This close partnership ensures that our research is not only scientifically rigorous but also designed to support the practical needs of forest managers and policymakers working to protect Peru's forests and the CITES-listed tree species they contain.
Out of the lab and into the forest
With support from the Grantham Amplification Fund, the FORGE team (including Xiaoqi Zhuang, Dr Robert Bryant and Dr Christopher Bousfield) visited Peru for 10 days to work closely with the project partners, OSINFOR, in Lima and Pucallpa, in the Peruvian Amazon.
The visit combined meetings with government agencies, researchers, and local stakeholders with a visit to the OSINFOR laboratory in Pucallpa, a tour of a certified timber processing facility and extensive fieldwork, allowing the team to observe how forest engineers conduct forest inventories.
“For me, the most valuable part of the trip was seeing first-hand how forest inventory data is collected under challenging field conditions,” says Jefersson. “As a computer scientist developing artificial intelligence tools, it is easy to focus on algorithms in the lab. Spending time with forest engineers in the Amazon highlighted the practical realities of data collection, from navigating dense rainforest to capturing reliable drone imagery and accurately recording tree measurements. This experience informed the development of improved data collection protocols and ensured that the AI tools we are developing address real operational needs rather than laboratory scenarios.”
Forestry management and the ARBOR plugin
One of the project's main outputs is ARBOR, a novel QGIS plugin that automatically detects Shihuahuaco trees from drone imagery. By integrating AI into a widely used geographic information system, ARBOR enables forestry professionals to analyse drone surveys more efficiently, reducing the time and effort required for manual interpretation.
“The drone data collected during our field visit to Peru allowed us to test and validate ARBOR under real operational conditions, providing valuable feedback for its continued development. Following further development and validation, we plan to make ARBOR available free of charge to researchers, conservation organisations, and forest authorities upon request, supporting wider adoption of AI-assisted approaches for tropical forest monitoring.”
How the Grantham Amplification Fund helped
The Grantham Amplification Fund not only enabled the team to conduct valuable fieldwork in Peru, but it also helped them recruit a Research Assistant, who played a key role in developing and evaluating the ARBOR QGIS plugin.
“More broadly, the Grantham Amplification Fund enabled FORGE to fulfil its role as a seed project. It strengthened our long-term partnership with OSINFOR, established the foundations for future collaborative research, and positioned us to pursue larger research projects focused on applying artificial intelligence to support sustainable forest management and biodiversity conservation.”
What is next for the project?
The FORGE project has laid the foundations for a much more ambitious research programme. The team has now been awarded an APEX Award from the Royal Society, in partnership with the British Academy and the Royal Academy of Engineering, to launch the two-year NextGen-FI project.
“NextGen-FI aims to transform tropical forest monitoring by combining drone-based remote sensing, advanced imaging technologies, and state-of-the-art artificial intelligence. We will develop new AI models capable of identifying individual tree species in some of the world's most biodiverse forests, including recognising rare and endangered species and detecting species that have never been encountered during training. This represents a significant advance over current approaches, which often struggle with the extraordinary diversity and complexity of tropical forest canopies.”
“Ultimately, our goal is to develop practical tools that help forest authorities monitor biodiversity, detect illegal logging, and support sustainable forest management. By building directly on the knowledge, partnerships, and data established through FORGE, NextGen-FI moves us closer to deploying AI technologies that can make a tangible contribution to the conservation of tropical forests and the protection of threatened tree species.”
The project was just one of 14 innovative sustainability research projects which received funding through the Grantham Amplification Fund Initiative.