Protecting bananas through AI
A new smartphone tool conceived for small banana farmers scans plants for signs of five major diseases and one common pest
A new smartphone tool made for banana farmers scans plants for signs of five major diseases and one common pest. Through the testing phase in Colombia, the Democratic Republic of the Congo, India, Benin, China, and Uganda, the tool provided a 90 percent successful detection rate.
Bananas: An endangered species?
The banana fruit remains as popular as ever, yet its crops across the globe have been hit with an infectious fungus.
The Cavendish species of banana, which was introduced in 1965, is the main banana export. It is being severely affected by Tropical Race 4, a fungal disease that began in Malaysia in 1990 and has since spread to Southeast Asia, Australia, and eventually Africa in 2013.
This is not the first time a fungus has wiped out an entire species of the bright yellow fruit. By 1965, the Gros Michel species of banana was eradicated after what was called the “Panama disease”, a different strain of a similar fungal disease wiped out commercial banana plantations around the world.
After such events, the industry looked for a new version of the crop, settling on the (commonly perceived) inferior Cavendish as its only alternative. It was then cloned and grown across the globe, making the single species (a monoculture) quite susceptible to spreading infection (once one plant gets hit with the fungus, they’re basically all in trouble).
The most significant challenge is that the fungus stays in the soil. It negatively affects the plant’s vascular system and prevents it from grabbing water from the ground. The only way to remove it is to burn the banana plantations to the ground, then begin fresh in a new location with a new species of banana crops. The disease is spreading because bad practices from the 1960s are still in place.
“It cannot be eradicated but it can be limited if a wide range of strong preventative and mitigation initiatives are put in place and rigorously implemented,” Joao Augusto, a plant pathologist told CNN in 2015. “In countries where the disease is endemic, the growers have learned to live with it.”
AI to the rescue
The new smartphone tool is built into an app named “Tumaini” (“hope” in Swahili). It was conceived by Bioversity International, a global research-for-development organization, to help small banana growers quickly detect a disease or pest and prevent a wide outbreak from happening. The app aims to connect them to extension workers to quickly stem the outbreak. It can also upload data to a global system for large-scale monitoring. The app’s goal is to facilitate an efficient and easily deployable response to support banana farmers in need of crop disease control.
“The overall high accuracy rates obtained while testing the beta version of the app show that Tumaini has what it takes to become a very useful early disease and pest detection tool,” stated Guy Blomme, a researcher at Bioversity International. “It has great potential for eventual integration into a fully automated mobile app that integrates drone and satellite imagery to help millions of banana farmers in low-income countries have just-in-time access to information on crop diseases.”
Beyond the app
Exponential improvements in image-recognition technology made the Tumaini app possible. In order to build such, researchers uploaded 20,000 images that depicted various visible banana disease and pest symptoms. With this information, the app scans photos of parts of the fruit, bunch, or plant to determine the nature of the disease or pest. It then provides the necessary steps to address the specific disease. Furthermore, the app also records the data, including geographic location, and feeds it into a larger database.
Existing crop disease detection models focus mainly on leaf symptoms and can only accurately function when pictures contain detached leaves on a plain background. The novelty in this app is that it can detect symptoms on any part of the crop, and is trained to read images of lower quality, including those with background noise, such as other plants or leaves, to maximize accuracy.
“This is not just an app,” said Michael Selvaraj, the lead author of the research. “But a tool that contributes to an early warning system that supports farmers directly, enabling better crop protection and development and decision making to address food security.”