By Mary Hearty

With the recurrent challenges facing the African food system such as climate change, pests and diseases, market failure and inefficient value chains, and adulterated agro-inputs, investing in innovations and technology alongside good agricultural practices and making them easily accessible to farmers is the greatest opportunity to transform the sector.

Artificial intelligence (AI) has emerged as one of the most potent solutions to the many challenges this sector faces, especially in sub-Saharan Africa where agriculture is the backbone of socio-economic development.

For instance, AI systems can help enhance productivity while minimizing environmental impacts; can support farmers in monitoring the condition of their crops, soil, and livestock and provide timely recommendations on particular actions and decisions, according to the Food and Agriculture Organization of the United Nations.

The Africa Centre for Technology Studies (ACTS) with support from the International Development Research Centre (IDRC) and the Swedish International Development Cooperation Agency (SIDA) is offering scholarship program for fostering the talent needed to meet a growing demand for research and development in responsible AI that can help promote sustainable agriculture in Africa.

The program is to support at least 12 scholars to undertake and successfully complete PhD research in AI and machine learning for a period of up to 36 months.

In addition, it aims to supports at least eight early career academics to strengthen R&D capacities in AI and machine learning for a period of at least 24 months; and also to facilitate professional development for the PhD and economic commission for Africa (ECA) scholarship holders.

Among these projects that ACTS is currently supporting include: locust management, stress detection on tomato plants, prediction of crop yield, preservation of crops, and also boosting dairy production.

Locust Management

On locust management, ACTS is supporting the development of an early warning system for the management of locust invasion in Zambia.

Locusts have contributed about 30-40% of the world’s food production. In 2020, the Food and Agriculture Organization of the United Nations (FAO) reported that 88,700 households in Zambia were affected as a result of the destruction of an estimated 47,000 hectares.

Therefore, the project will identify challenges faced by farmers and government in controlling locust invasion using existing early warning systems; assess how AI can be used to detect locust invasion; design an early warning system for locust invasion based on AI and other emerging technologies and evaluate a suitable early warning system model.

This is according to Brian Halubanza, a PhD candidate in Computer Science at the University of Zambia who is steering this project during his presentation at the 3rd Prof Calestous Juma Lecture Series.

Stress Detection on Tomato

Another project aims to address food security through a system used to detect stress on tomato plants under climate and infection-based simulated environments.

Ariane Houetohossou, a PhD student at the University of Abomey-Calavi (UAC) in Benin, who is conducting this project said it uses deep learning (DP) technique, a method inspired by the functioning of the human neurons for the early identification of a highly destructive pest to tomato plants and fruits known as Tuta absoluta.

Diseases remain one of the most critical factors contributing to the considerable reduction of agricultural yields. In tomatoes for instance, they can reduce the output up to zero, hence early detection is essential in plat monitoring.

Prediction of Crop Yields

Again, ACTS is supporting a project that contributes to food security through optimization of machine learning techniques and performances to predict the yield of maize culture under controlled weather and fertilization patterns.

This is because agricultural production conditions in West Africa are becoming more adverse due to climate change. Using AI methods and relevant technologies could help model growth and predict the yield of maize, according to Souand Peace Gloria, a PhD student at the University of Abomey-Calavi in Benin, who is piloting the project.

The research intends to perform critical reviews on machine learning techniques used for cereal yield prediction; determine associated pre-generated weather characteristics using maize yield data cultivated in real environments; assess the impact of pre-generated weather characteristics and fertilization levels on the growth parameters and the yield of maize; and assess the performance of machine learning techniques for maize yield prediction and compute the final optimization technique of maize.

Preservation of Crops

ACTS is supporting the development of a “smart granary” that will allow farmers to preserve their fruits, vegetables and cereals. The smart granary is made of automatic ventilation system, and automatic watering systems that are triggered through a solar system as energy to cool these products when the temperature is unfavorable.

The technology also features a cloud platform with AI system for monitoring products with key performance indicators on evolution of production. The cloud platform with AI system has sensors that collect data such as temperature.

Majority of African produce lose up to 70% of the production due to pests and excessive heat. In East Africa for instance, FAO reports that economic loses in the dairy sector due spoilage and waste could average as much as USD 90 million per year.

Ndeye Fatou, Big Data Engineer and AI specialist in Senegal, and the one who pilots this project stated that this smart technology is meant to increase profitability, promotes affordable selling of prices for fruits and vegetables, and modernization and automation of agriculture.

In addition, it can help decrease the number of deaths due to poisoning from various products; and eradicate of famine and poverty as well as malnutrition. The environmental benefits are; the smart technology promotes sustainability of products as it uses renewable source of energy; preservation of the agricultural balance; and protection of the environment against waste clutter.

“We are planning to deploy the device with 100 large scale Senegalese producers within six months, and 1000 large scale producers across 10 African countries,” Fatou revealed.

She also plans to develop systems that trigger pesticides and have drones for catching and processing to prevent parasitic infections of plants; and to make the system intelligent without using connected objects through emerging technologies such as machine learning, big data and AI.

Dairy Productivity

Another project being supported by the ACTS aims to increase milk yields through intelligent agents modelling system, individual farmers’ knowledge, learning and sharing experience in a farmer-to-farmer network

Dr Devotha Nyambo, a PhD candidate at the Nelson Mandela African Institution of Science and Technology said some of the models the project places emphasis on are data sets, mobile app for farmer-to-farmer learning, agent-based model for farmer-to-farmer learning, and mobile app for learning recommendations.

She noted that in attempts to increase milk yields, smallholder dairy farmers engage in various improved husbandry practices such as breeding technologies, health management and watering frequencies among others.

Nevertheless, with insufficient extension services, farmers are stuck in failure loops and unsuccessful attempts that may undermine desire to continue farming.

She said simulation results indicate that within a period of 290 days, chances are high that almost all the farmers will adopt the agent based model to improve their milk production.

Besides giving programs, ACTS is also trying to build a network of people who have an interest to develop the application of AI and machine learning in sustainable development areas in SSA.

These networks are certain complementary activities around the scholarship program. The beneficiaries to these networks consist of – PhD students and Post-Docs, supervisors, partner ethical AI experts, other scholars in the beneficiary universities, among others.