Uganda's PlantVillage team developing AI model for pest and disease control in banana

Posted on

Globally, banana is the fourth most important crop on the food market after rice, wheat, and maize. In many developing and developed countries, banana is an important crop, and here in Uganda, it is served on every table as a staple meal.

The crop is used in a variety of dishes, including green boiled or fried bananas, desserts like sukalindizi and bogoya, and roasted bananas like gonja. It can also be added value to make biscuits and cakes for sale.

Nutritionally, bananas are known for being rich in potassium, fiber, vitamins C and B6, and magnesium.

Western Uganda leads in banana production, while the Central, Eastern, and Northern regions are the least producers.

Advancing the Banana AI Model for Pest and Disease Control

Banana, just like coffee, which is also widely grown in Uganda, faces a severe threat of pests and diseases, posing the biggest challenge to production.

Diseases like banana bacterial wilt completely ravage the farm and may lead to a 100 percent loss of yield.

PlantVillage, is working on a sustainable solution to the problem, and soon a farmer will just need a smartphone to find instant assistance rather than the traditional way of contacting extension services which may be far from reach.

The recent release of the coffee AI model, which is being improved to be able to detect more pests and diseases, has proved very effective in combating pests and diseases in coffee and inspired the engineering of the banana AI.

Banana xanthomonas wilt in banana

Just like how it was done in advancing the coffee AI module, PlantVillage’s  embarked on collecting images of both diseased and healthy banana leaves, male fruit buds, and pseudo stems that will be used to put up the banana AI model.

"As a field officer, my farmers will be able to differentiate and easily identify banana diseases and pests in their fields, thereby controlling them. This will boost banana production in the country and the world at large," affirmed Promise Owamaani, a PlantVillage field research and extension officer in Mpigi District, Central Uganda.

Richard Ssekamatte, field officer from Masaka District in Uganda, takes images for the banana AI model

The process of engineering the model is gradual and includes, apart from collecting different images from the field, annotating the images, AI field trials to ensure the model can recognize different diseases and pests.

The image collection phase has already been accomplished, and the annotation phase and field tests are ongoing. The model will be launched soon for trials by field officers for hands-on experience and accuracy.

The banana AI is designed to detect black sigatoka, yellow sigatoka on leaves, banana xanthomonas wilt (also known as banana bacterial wilt) on leaves, fruits, and pseudo stems, banana weevil on pseudo stems, and healthy plant parts.

Black Sigatoka disease in bananas

The AI model will be able to detect and diagnose all these diseases and pests for the farmer and give immediate advice on control measures. At this point, every farmer will have found a lasting solution, and banana production in the country will receive a major boost.

"Diseases like banana bacterial wilt have led to big losses here in Masaka. As farmers, our efforts on these plantations have yielded no positive results. We hope this banana AI will solve the problem," said Kimera Deus, a lead farmer in Masaka District, Central Uganda.

  • Written by Agnes Kampire 
Heart Heart icon