AI Maps Benin’s Cashew for the First Time: A Major Step towards Sustainable Production

Cotonou, Benin (July 26th, 2023) – A new study has shown that remote sensing and machine learning can be used to accurately map smallholder tree crops in emerging economies. Researchers from the University of Minnesota and partners including TechnoServe have published research in the September 2023 edition of the journal Remote Sensing of Environment detailing this groundbreaking work for Benin’s cashew sector. 

The research in Benin was carried out as part of the U.S. Department of Agriculture BeninCajù project, which is led by TechnoServe and has benefitted more than 120,000 people in the country.

TechnoServe and students from Carnegie Mellon University-Africa have turned the results of this research – detailed information about where Benin’s cashews are grown – into a dashboard for policymakers, development organizations, and the private sector to efficiently target training, cashew tree nurseries, and other resources that benefit the livelihoods of smallholder farmers. 

Benin is the world’s tenth-leading producer of raw cashew nuts in the world, and the crop is an important part of the national economy, contributing 15% of export earnings and providing livelihoods to some 200,000 farmers. Nevertheless, there was no comprehensive map of where the country’s cashew farms were located, making it difficult for stakeholders to determine where to provide extension services or start plant nurseries, or to track whether farming is occurring in protected areas. Geospatial tools can provide a cost efficient way of mapping smallholder crops like cashew at scale. 

Over the course of three years, the researchers created and trained a machine-learning algorithm to analyze satellite imagery in order to identify the presence and density of cashew farms. While this technology has found increasing use for large farms producing row crops such as maize, rice  and soy, its application to small parcels of tree crops like cashew had not been successfully done at scale.

The researchers used Planet’s 2.4-m, 4-band Basemaps and Airbus’ 0.5-m Pléiades satellite imagery of Benin, and then applied a machine learning algorithm to identify the presence of cashew farms and determine if these were high-  or low-density. By comparing these results to a large set of ground-truthing samples collected by field teams (primarily from the BeninCajù program) , it was found that the algorithms were 85% accurate in identifying the presence of cashew trees.

The study will help target training and other resources so that Benin can meet its ambitious national targets of increasing cashew production by 100,000 MT to 300,000 MT in 2026. Cashew production doubled to approximately 200,000 MT in 2021, and the study confirmed official figures that area planted with cashew nearly doubled between 2015 and 2021. Going forward, the study identified the potential to increase production through intensification and targeted thinning and pruning.

“Satellite remote sensing technology, particularly when combined with the prowess of AI and machine learning, presents a transformative opportunity for the developing world to guide the sustainable expansion of smallholder tree crops,” said Zhenong Jin, Ph.D., assistant professor in the Department of Bioproducts and Biosystems Engineering at the University of Minnesota.

“This initiative demonstrates the potential to use geospatial research and tools to improve the livelihoods of smallholder farmers in Benin and across Africa,” said Dave Hale, director of TechnoServe Labs, the organization’s team based in Kigali, Rwanda and Silicon Valley dedicated to identifying, testing, and implementing promising technologies to deliver market-based solutions to poverty on a global scale. “This applied research is a great example of TechnoServe Labs’ work with  the academic community—including agricultural research leaders at the University of Minnesota and tech innovators at Carnegie Mellon University – Africa in Rwanda and African technical schools such as EpiTech in Benin.”

“Cashew represents an enormous opportunity for Benin and Benin’s farmers,” said Epitace Nobera, chief of party for BeninCajù. “Having accurate, timely data about where cashews are grown and in what density helps ensure that those farmers have the information, resources, and services they need to succeed and lift their families out of poverty.”

The researchers from the University of Minnesota and TechnoServe will also apply the approach in support of a large-scale cashew project based in Abidjan, Cote d’Ivoire.

## 

About TechnoServe

Founded in 1968, TechnoServe is a leader in harnessing the power of the private sector to help people lift themselves out of poverty for good. A non-profit organization working in 30 countries, we work with people to build a better future through regenerative farms, businesses, and markets that increase incomes. Our vision is a sustainable world where all people in low-income communities have the opportunity to prosper.

More information at: Twitter @TechnoServe | Facebook @TechnoServe | LinkedIn @TechnoServe

About the University of Minnesota College of Food, Agricultural and Natural Resource Sciences

The University of Minnesota’s College of Food, Agricultural and Natural Resource Sciences (CFANS) strives to inspire minds, nourish people, and sustainably enhance the natural environment. CFANS has a legacy of innovation, bringing discoveries to life through science and educating the next generation of leaders. Every day, students, faculty, and researchers use science to address the grand challenges of the world today and in the future. CFANS offers an unparalleled expanse of experiential learning opportunities for students and the community, with 12 academic departments, 10 research and outreach centers across the state, the Minnesota Landscape Arboretum, the Bell Museum of Natural History, and dozens of interdisciplinary centers.