Using Artificial Intelligence to Protect the World’s Banana Crops

To further elaborate on our conversation about the use of artificial intelligence (AI) in the food industry earlier this week, I found a study about how a phone application called Tumaini is using AI to help farmers protect their fruits. This tool boasts an astounding average success rate of 90 percent in detecting pests and diseases that infect and potentially destroy banana crops. This highlights the reality that even in remote corners of the world, tools that utilize the power of artificial intelligence are becoming not only accessible, but a necessity. In addition to pest and disease detection, farmers are able to more efficiently manage their farms with linkages to markets, extension workers, satellite images, and information about their crops.

Study Information:

Michael Gomez Selvaraj, Alejandro Vergara, Henry Ruiz, Nancy Safari, Sivalingam Elayabalan, Walter Ocimati, Guy Blomme. AI-powered banana diseases and pest detection. Plant Methods, 2019; 15 (1) DOI: 10.1186/s13007-01

What? An application for smartphones was developed that currently scans plants for signs of five major disease and one common pest.

What?  This tool provides farmers in low-income countries an opportunity to improve surveillance of their crops, allowing for enhanced efficiency in control and mitigation response to pests and crop diseases which can reduce the number of crops and production lost. The intent of this work is to ultimately combine with the use of drones to create a satellite-powered, globally connected network which will help to control outbreaks of pests and diseases.

Why is this important? Food is an enormous concern, especially in developing countries. Bananas were targeted because they are the world’s most popular fruit, and a dietary staple of many families. A loss in crop production not only limits the supply available and drives up cost, but it can be catastrophic for the small farmers that base their livelihood on the sale of their banana crops. One fungus, known as the Fusarium Tropical Race 4 fungus, has generated banana crop losses totaling an estimated $121 Million (all amounts in US Dollars) in Indonesia, $253 Million in Taiwan and $14.1 million in Malaysia, according to Aquino, Bandolas, and Lim (2013).

Where did the study occur? Columbia, the Democratic Republic of the Congo, India, Benin, China, and Uganda.

How did they do it? Researcher uploaded 20,000 images that depicted various banana disease and pest symptoms to create an index where, when the fruit, bunch, or plant is scanned, it matches the photos to determine the nature of the issue. It then provides the steps the farmer needs to take to address the issue. All of this data is collected by the application, including geographic location, and fed into a larger collective database.

Why is this better than what is currently available? Current models focus on leaf symptoms and require precise conditions, such a detached leaf on a plain background. This program can detect symptoms in any part of the crop, even in situations were the image quality is poor, there is excessive background noise, like other plants or leaves.

Who did this? This study was made possible through collaboration between Biodiversity International and the International Center for Tropical Agriculture (CIAT), which has previously demonstrated the potential of artificial intelligence, IoT (Internet of Things), robotics, satellites, cloud computing, and machine learning for agricultural benefit.