Author Archives: Thiago Gouvea

Transforming Biodiversity Monitoring in the Amazon: Participation in the AmazonIA – Mamirauá Meeting

At the AmazonIA – Mamirauá Meeting, held in the heart of the Amazon, international experts and institutions joined forces to define a new agenda for biodiversity monitoring. Our group contributed perspectives from interactive machine learning and large-scale acoustic monitoring—continuing the mission that earned us a top-3 result in the XPRIZE Rainforest competition.

Thiago Gouvêa, CST group leader, joins other participants of the AmazonIA – Mamirauá Meeting in visiting a riverside family engaged in sustainable honey production in the Amazon.


From 8 to 10 September 2025, the Mamirauá Institute for Sustainable Development hosted the AmazonIA – Mamirauá Meeting in Tefé (Amazonas, Brazil)—a landmark gathering uniting scientists, conservation practitioners, technologists, and philanthropic organizations to shape the future of biodiversity monitoring in the Amazon. The meeting culminated in the Mamirauá Declaration, a joint commitment by more than thirty institutions to develop and deploy advanced technologies for long-term conservation of the Amazon rainforest and its peoples.

The event’s focus on integrating remote sensing, bioacoustics, and artificial intelligence resonated strongly with our own research on interactive machine learning for wildlife monitoring. Our work aims to bridge automation and expert knowledge in the analysis of large, unlabeled acoustic datasets, improving the detection and classification of animal vocalizations while keeping humans meaningfully involved in the learning process. These methods form the foundation of our ongoing development of open, collaborative tools for ecological monitoring—approaches that have been recognized internationally, including our Top-3 placement in the $10 M XPRIZE Rainforest competition.

The Mamirauá Declaration, drafted at the conclusion of the event, marks a significant step toward a shared international effort to monitor biodiversity and climate resilience in the Amazon. By linking scientific innovation with local stewardship, the initiative lays the groundwork for cross-border cooperation ahead of COP30 in Belém. Our group is honored to contribute to this collective endeavor, working alongside Brazilian and international partners to develop transparent, adaptive, and community-oriented AI tools for biodiversity observation at scale.

IML Solutions for Effective Wildlife Monitoring to be Featured at IJCAI

We, a collaborative team of machine learning scientists and ecologists entrusted with the management of biodiversity resources, are leveraging the power of interactive machine learning to drive impactful conservation strategies. Our endeavor will be published at the prestigious International Joint Conference on Artificial Intelligence (IJCAI).

The alarming acceleration of biodiversity loss demands innovative solutions and urgent action. At the forefront of this battle, biosphere reserves have emerged as crucial sites of global significance, harboring rich biodiversity and offering sustainable development opportunities. The management of biosphere reserves demands adaptive strategies that address the ongoing threats to biodiversity. However, the lack of comprehensive information on the status and trends of wildlife populations hinders effective decision-making. To bridge this knowledge gap, the research team delved into the realm of Passive Acoustic Monitoring (PAM).

Passive Acoustic Monitoring, a scalable and reproducible method for wildlife monitoring, has emerged as a promising avenue for closing the information gap in biodiversity conservation. By leveraging sound recordings, PAM enables continuous, non-invasive monitoring of animal wildlife across diverse ecosystems. Yet, the adoption of PAM has been hindered by challenges in data management and analysis, limiting its broader implementation by agencies with a mandate for biodiversity management. Recognizing the need for automation and streamlined analysis, our interdisciplinary team is developping state-of-the-art interactive machine learning tools tailored specifically for PAM data.

Our upcoming article, to be featured in the AI and Social Good section of the IJCAI conference proceedings, showcases research integrating advanced techniques like active learning and transfer learning. These techniques enable domain experts to enhance their work, a valuable and limited resource, by utilizing existing knowledge from databases to address specific and potentially novel scenarios encountered in different biosphere reserves. Complementing these advancements, our tools incorporate intuitive data visualization and intelligent user interfaces, enhancing the system’s usability and accessibility. This holistic approach equips conservation agencies with the necessary means to make data-driven decisions and drive effective conservation strategies.

Read more

H. Kath, T. S. Gouvêa, and D. Sonntag, ‘A Human-in-the-Loop Tool for Annotating Passive Acoustic Monitoring Datasets’, in Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, Macau, SAR China: International Joint Conferences on Artificial Intelligence Organization, Aug. 2023, pp. 7140–7144. doi: 10.24963/ijcai.2023/835.

T. S. Gouvêa et al., ‘Interactive Machine Learning Solutions for Acoustic Monitoring of Animal Wildlife in Biosphere Reserves’, in Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, Macau, SAR China: International Joint Conferences on Artificial Intelligence Organization, Aug. 2023, pp. 6405–6413. doi: 10.24963/ijcai.2023/711.