PlantVillage at Penn State is an artificial intelligence (AI)-first software solution used by the Food and Agriculture Organization of the United Nations (FAO) to tackle multiple transboundary pests like the desert locust, fall armyworm and red palm weevil.

With support from USAID via the Current and Emerging Threats to Crops Innovation Lab (CETC IL), the Bill and Melinda Gates Foundation and private donors, PlantVillage leverages its unique experience to fight the emerging threat of banana bunchy top virus (BBTV) in Tanzania.

“The challenge for Tanzania is that extension workers lack awareness of BBTV. At the same time, the central authority, such as the Tanzania Plant Health and Pesticide Authority (TPHPA), need real-time updates on the disease so they can direct national responses,” said David Hughes, Director of CETC IL and PlantVillage at Penn State.

Integrating AI technology

The solution involves an integrated platform featuring an AI assistant within the PlantVillage app, capable of diagnosing infected plants offline in the field. Utilizing TensorFlow software from Google, with which PlantVillage has a longstanding relationship, this technology enables prompt detection of BBTV.

Dr. Hamad Lyimo, principal agricultural officer at the TPHPA, emphasized that “to conduct effective surveillance in Tanzania, a country with a total area of 945,087 square kilometers demands huge resources and time, which is always scarce. This necessitates the adoption of new technologies that plant protection officers and extension officers can accurately use at the village level and transfer data collected in time and at the minimum cost.”

The new model was created with the help of the PlantVillage ground team in Tanzania, Uganda, Kenya, Ethiopia, Malawi and Mozambique. The team used PlantVillage AI to monitor and report in real-time the incidence of disease, which is an essential first step in controlling BBTV. The system obtains a broad selection of images of banana plants, including healthy ones, those affected by BBTV and other diseases that affect bananas.

While the current version of the model is not perfect, it was released to urgently address the need for extension services support. Continuous updates will be made to improve its performance.

In addition to the AI model, a novel mapping tool has also been developed. This tool enables experts to review and update records.

Field officers can now pinpoint their location, enhancing survey efficiency and effectiveness.

A recent discovery by one of the PlantVillage officers, Ashura Swai, on February 19 underscores the importance of these efforts. Swai identified a previously unknown infected location, enabling swift communication with local authorities for necessary action.

“As I conducted surveys, I discovered BBTV-infected plants in Mutara Village, Muzye Ward, Kasulu District, Kigoma Region. When I inquired with district plant protection officer (DPPO), Mr. Amir Kasali, about BBTV in Kasulu District during their surveys, he indicated that they had not found any. However, I informed him of the newly identified BBTV area. Now, I eagerly await their response and discussion with other experts on the next steps,” noted Ashura.

Utilizing the PlantVillage AI Observatory

The PlantVillage AI Observatory provides a platform for local extension services and headquarters to monitor the accuracy of the generated records in real-time. This helps them to make informed decisions and take necessary actions. In addition, the Observatory enables government officials to communicate with the ground staff and access other relevant data collected. This, in turn, facilitates better decision-making and enhances the overall efficiency of the system.

“PlantVillage artificial intelligence (AI) app and Observatory give answers to these challenges. With this technology, surveys are done and data transferred in near real-time for processing and informing timely decisions. It is now important that this technology is delivered to plant protection officers and extension officers all over the country,” added Dr. Lyimo, TPHPA.

PlantVillage is also collaborating with the International Institute of Tropical Agriculture (IITA) via a subaward from CETC IL to integrate virus testing using point-of-care diagnostics as a means of authenticating virus occurrence into its system.

“PlantVillage AI is set to enhance BBTV surveillance by extension officials and trained public. IITA is engaged in rapid diagnostics confirmation of BBTV occurrence using an isothermal recombinase-polymerase amplification (RPA) assay. Together, these tools form a foolproof assessment and contain the BBTV epidemic in East Africa,” explained Lava Kumar, IITA.

As Tanzania’s battle against BBTV intensifies, leveraging cutting-edge AI technology and collaborative partnerships is vital to protecting banana crops and promoting food security.