University of Coimbra develops AI system for environmental monitoring in agriculture

A team of researchers from the Faculty of Sciences and Technology at the University of Coimbra is developing a system based on artificial intelligence (AI) and machine learning (ML) to monitor ecological footprints in agriculture.

The Pegada 4.0 project, led by the University of Évora and funded by the Recovery and Resilience Plan (PRR), focuses on five key indicators: carbon dioxide, water resources, diffuse pollution, landscape and biodiversity. It aims to promote more sustainable production by measuring and reducing the environmental impact of agricultural practices.

Catarina Silva, professor at the Department of Informatics Engineering (DEI) and researcher at the Centre for Informatics and Systems of the University of Coimbra (CISUC), explains that FCTUC's main focus is on the biodiversity footprint, fostering its conservation and growth in the agricultural plots defined in the project. "We are working on the monitoring of biodiversity, the collection and processing of information on the species present and their evolution over time," says the project leader at the UC.

"With this innovative approach, we expect agriculture in Portugal to become more efficient and environmentally friendly, setting a new standard for sustainable agricultural management," the team concludes.

The methodology used involves collecting multimodal data, including measurements of temperature and humidity, pictures of insects and plants, sound recordings of birds and more. The goal is to integrate and analyse this information to automatically identify species and discover new ones over time using dynamic AI models.

The project involves 20 agricultural partners and several farms that actively participate in environmental monitoring, and also Agroinsider, a company linked to the University of Évora.

Dinis Costa, a DEI student involved in the research, highlights the importance of "SmartAg", an application developed within the project that allows farmers to send images and sounds in real time for scientific analysis. "The application is used to record evidence of species, with information on the species, coordinates and time of recording," he explains.

"The project represents a major step forward in precision agriculture, which relies on dynamic AI and machine learning models to improve efficiency," says DEI professor and CISUC researcher Bernardete Ribeiro, adding, "We want to implement these models on low-cost hardware, making the process more effective, from design to implementation in the field."

Footprint monitoring also includes installing insect traps that capture images for species identification. According to the DEI student, "'SmartAg' allows farmers to record the species they observe. These images are automatically classified using intelligent models. In the initial phase, farmers will only need to validate that the AI classification is correct, which will help continuous improvement of the models," he explains.

The project will also explore the impact of landscape changes on biodiversity and demonstrate how sustainable management can benefit the agricultural ecosystem. With the AI models already field-tested, the FCTUC team is working to integrate these technologies into agricultural practices.

"With this innovative approach, we expect agriculture in Portugal to become more efficient and environmentally friendly, setting a new standard for sustainable agricultural management," the team concludes.