Detecting early signs of banana diseases

New technology could help protect the world’s most important fruit crop, the banana. A smartphone app developed by scientists on three continents can detect early signs of diseases that affect and eventually decimate banana trees, according to a paper published in the journal Plant Methods.

In field studies in India, the Democratic Republic of Congo, Benin, Colombia and Uganda, it has been shown to be more than 90% accurate overall in identifying fungal, viral and bacterial diseases. 

Bananas are a staple food throughout Asia, Africa and Latin America, with around 145 million tonnes grown each year. 

Existing disease-detecting techniques mostly rely on identifying symptoms on leaves, missing out on other disease-affected parts such as the fruit or pseudostem.  They also don’t identify the diseases and pests in real time.  

To find a better alternative, a team led by Michael Gomez Selvaraj from the International Centre for Tropical Agriculture in Colombia, developed a computer model using a deep learning algorithm then trained it with 18,000 images of disease- and pest-affected banana plants from farms in India and Africa. 

When tested in fields, the app was able to detect the difference between healthy and infected plant parts for diseases such as banana bacterial wilt, the lethal fungal infection Panama disease, leaf-spot fungal diseases, bunchy top viral disease and the corm weevil pest.

The researchers say the app would allow for early identification of pest and diseases and their management in banana farms using a simple hand-held device.

Next, they plan to devise a model and an application for smartphone devices for detecting diseases in other crops such as common bean, cassava, potato and sweet potato.