The app can detect Fusarium wilt, Xanthomonas wilt, bunchy top disease, black sigatoka, yellow sigatoka, and corm weevil.
Fusarium Tropical race 4 fungus (TR4) has decimated banana plantations and smallholders’ crops in Asia and Africa and has now spread to Latin America. Last week, Colombian officials officially confirmed the presence of TR4 in La Guajira province, declaring a state of national emergency as a result.
Developed with support from Bioversity International and the International Center for Tropical Agriculture (CIAT), the AI-powered tool is built into an app called Tumaini – Swahili for ‘hope’ – that allows farmers to take action quickly, thus preventing a widespread outbreak.
The information is also uploaded to a global system that allows for large-scale monitoring.
The app’s efficacy is down to huge improvements in image-recognition technology. While existing disease detection instruments can only provide accurate results using a photograph of a diseased leaf on a plain background, Tumaini can decipher the required information using a photograph of any part of the plant – fruit, bunch or leaves. It can also process lower-quality images and can isolate the banana plant from other plants in the background.
Farmers use the smartphone app to scan their crops for signs of the disease or pest. The model is linked to a database which stores 20,000 original pre-screened banana images collected on real farmers’ fields in Africa, Latin America, and South India.
Experimental results using the model were accurate between 70 and 99% of the time and the model system could easily be used for other crops, write the researchers.
"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," said Guy Blomme, from 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."
'A significant high success rate'
CIGIAR researchers have already planned to use it to recognize diseases in crops including the common bean, cassava, potato, and sweet potato.
“We are hoping to achieve a valuable impact on sustainable development and strengthen banana value chains,” they write in a paper detailing the study, published in Plant Methods.
"This is not just an app," said Michael Selvaraj, lead author, who developed the tool with colleagues from Bioversity International in Africa, "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."
Farmers can easily use the detection model, based on three different convolutional neural network (CNN) architectures using a transfer learning approach, making it a “robust” digital strategy to detect banana disease and pests.
“This significant high success rate makes the model a useful early disease and pest detection tool, and this research could be further extended to develop a fully automated mobile app to help millions of banana farmers in developing countries.”
According to some estimates, the disease caused losses of US$121 million in Indonesia, US$253.3m in Taiwan, and US$14.1m in Malaysia (Aquino, Bandoles and Lim, 2013).
Source: Plant Methods
Available online ahead of print, doi.org/10.1186/s13007-019-0475-z
"AI-powered banana diseases and pest detection"
Authors: Selvaraj et al.
