ESALQ/USP / FEALQ
Life Sciences
Although its origins are directly linked to teaching and research in the agricultural field, the Luiz de Queiroz School of Agriculture (ESALQ) has begun to develop innovative research in the environmental field, focused on the study of natural ecosystems, their biodiversity and dynamics, and environmental restoration and adaptation to agricultural landscapes. It includes laboratories such as LERF (Laboratory of Ecology and Forest Restoration) and LASTROP (Laboratory of Tropical Forestry), in addition to the ESA Herbarium, one of the ten largest herbariums in Brazil in terms of representation of tropical forests.
FEALQ, a partner institution in this initiative, manages Research, Extension, and Innovation, promoting teaching and knowledge. It is a private, non-profit entity responsible for the administrative and financial management of projects developed between members of universities and public and private institutions, both national and foreign.
Projects
The Brazilian Team was one of the winners of XPRIZE Rainforest, a five-year global scientific competition focused on diagnosing the biodiversity of tropical forests through innovative technologies, artificial intelligence, and automation. The challenge was to identify the largest number of species of flora and fauna in a 100-hectare area of tropical forest. The competition involved more than 400 teams from around the world, with the Brazilian Team being the only finalist from the Southern Hemisphere. The semi-final took place in Singapore and the final in the Brazilian Amazon Rainforest.
Led by Prof. Vinicius Castro Souza, from ESALQ/USP, the team brought together around 100 researchers, mostly Brazilian, from several renowned institutions, such as USP, UFSCar, and INPA. Advanced technologies, such as robots and autonomous drones, were used to remotely collect data on sounds, images, DNA samples, temperature, humidity, and other environmental parameters, which were analyzed using artificial intelligence and other innovative techniques. In both the semifinals and the finals, the Brazilian Team was able to identify more than 250 species in these extremely complex and little-known environments in just 24 hours of autonomous or remote work and 48 hours of analysis. The Brazilian Team’s proposal also included the use of deep learning to segment and identify Amazonian tree canopies.
The main limitation to these numbers being even higher is the scarcity of databases in tropical forests. As a result, even after the award, the team continues to refine its technological tools and produce libraries of sounds, images, and DNA. In an aerial mapping with drones covering 1,700 hectares of the Amazon, botanists identified and registered more than 250 tree species. With the images captured, it will be possible to refine deep learning models to locate and monitor species, allowing for a better understanding of local biodiversity and its dynamics over the years.
