Pest and disease management for food security – there’s an app for that! (and it uses artificial intelligence)
It’s fair to say that pest management has come a long way since the days of scarecrows!
The most recent green revolution wave focuses on how artificial intelligence can be used to boost food security, particularly through pest and weed management. The latest technologies use ‘deep learning’ – a process part of the broader terminology of machine learning whereby programmers go beyond just telling the computer what to do with task-specific algorithms and code, but instead develop multi-level, non-linear processing units that recognise patterns in large sets of data. The learning can be supervised (a programmer trains the computer), partially supervised or unsupervised (learns independently). Deep learning processes are already in use in many industries around the world, including agriculture. Application of AI to the agricultural sector can have enormous benefits for farmers and food production globally, in both developed and developing countries.
A key challenge to farmers in many developing regions of the world is accurately identify crop diseases, and a key challenge for researchers and extension agents is having the tools to assist. Diseases such as the cassava brown streak disease and cassava mosaic disease, could impact over 30 million farmers in East and Central Africa, impacting both security of income, food and nutrition. Other common diseases in this region include fungal diseases in banana crops such as Banana bunchy top virus, and late blight devastates potato farmers. New technologies using AI, big-data and mobile phones can take on the role of an extension worker in regions where it may not otherwise be possible for advanced scientific knowledge to reach the ground.
This week, we’ll take a look at 3 apps which are leading the way in using artificial intelligence to improve food systems and food security globally.
A research team from the CGIAR Research Program on Roots, Tubers and Bananas (RTB) has developed a mobile phone app to enable cassava farmers to instantly diagnose crop diseases by taking a photo. The app will allow plant diseases to be diagnosed, alerts to be sent to surrounding farmers, and will connect farmers to a network of extension officers. Through the app, farmers can take a photo of a diseased plant and be given an instant diagnosis.
James Legg from the International Institute of Tropical Agriculture (IITA), has said “Smallholders or extension officers with a basic smartphone with a camera will be able to download the app for free, fire it up, point it at a leaf with disease symptoms and get an instant diagnosis. That is truly revolutionary!”
To develop the app, the research team generated over 200,000 images of diseased crops in coastal Tanzania and Western Kenya, using cameras, spectrophotometers and drones. The images are used to develop AI algorithms which enable the diseases to be automatically recognised by the app.
David Hughes, associate professor of entomology and biology at Penn State University has said “The app employs AI in real time so the farmer can be an active participant in disease diagnosis and crop health management, leading to more yields for smallholder farmers”.
At the moment, the app is only programmed for cassava, but the team are now looking at expanding to other root, tuber and banana crops. The app aims to reach 35,000 farmers by July 2018.
Plantix is a similar app for automated disease diagnosis, developed in India by the Progressive Environmental & Agricultural Technologies (PEAT) in collaboration with the International Crop Research Institute for the Semi-Arid Tropics (ICRISAT) and the Acharya N G Ranga Agricultural University. The app enables farmers to take photos of their crops to receive critical information on symptoms, triggers, chemicals as well as treatments, through artificial intelligence algorithms and image recognition technology. The images are geo-tagged to enable real-time monitoring, inclusion of weather information, and to develop a meta-data base on the spatial distribution of crops and pests in the form of high resolution maps.
The Plantix database currently has over half a million images! This includes over 30 crops, and 120 crop diseases. This includes Fusarium wilt, Sterility mosaic disease and Phytophthora blight.
Dr David Bergvinson (Director General, ICRISAT), has said “The app could prove to be a game changer in the field, providing farmers free, reliable and quick diagnosis of crop damage. The simplified dashboard with easy-to-use features helps the app take on the role of an extension worker as well. The app is a novel experiment in using digital technologies for agriculture”.
The program has been tested in the Krishna district in Andhra Pradesh, and uses Indian regional languages of Telugu and Hindi, and will soon be extended to include other regional languages.
WeedID is a new app, developed by Masters student Yehezkiel Henson and a team of researchers at the University of Sydney, to assist in identifying weeds in rice fields in Cambodia. The app contains a photo dictionary, so when farmers observe a weed they can identify both the weed and the most appropriate method to manage it. The app aims to prevent the spread of weeds, and improve yields, incomes and knowledge in the region.
The developer, Yehezkiel Henson , has said: “WeedID contains a photo dictionary of the most common weeds in northwest Cambodian rice fields at different stages of growth. The app has images of seeds, seedlings, mature plants and flowers that will all help to identify the weeds which are devastating Cambodian farmers’ rice crops”.
The apps database now covers 45 weeds – which is 95% of the most frequent weeds identified in Cambodian rice fields. Weeds are a problem for rice production at various stages of the production process: weeds during cultivation significantly impact on yields (more than 50%), whilst contamination at harvest reduces the quality and consequently the price the rice can be sold for.
“The WeedID app contains links to specific management information and details the most appropriate way to manage the weed”, said Yehezkiel Henson. “Different weeds require different management techniques. Depending on the life cycle, nutrient requirements or mode of reproduction, we will employ a different method of management… Some weeds, such as awnless barnyard grass (Echinochloa colona) can be managed simply by flooding the fields to drown them. Sometimes we may need to apply specific selective herbicides to manage them, as in the case of the smallflower umbrella plant (Cyperus difformis), also known as ‘dirty Dora’ in Australia” said Yehezkiel.
The app has been commissioned as part of an ACIAR project examining sustainable intensification and diversification in the lowland rice system in northwest Cambodia, and developed from funding from International Environmental Weeds Foundation (IEWF) and the Crawford Fund.
Daniel Tan, from the University of Sydney, has said: “The Cambodian smallholder farmers we work with are also hungry for information and use the technology available to solve problems. Farmers have shown us photos of pest animals, weeds and diseased plants taken on their phones, which our team have been able to identify. This prompted the idea to develop a free, user-friendly mobile tool to identify weeds in rice paddies so farmers can manage the problem and improve the crops that support their families”.
Seems like a very apt solution!
We have already seen the huge success stories coming from using mobile phones to assist farmers in developing countries to access big data. The use of artificial intelligence in these apps and technologies is the next step to overcoming the challenges faced by farmers in developing regions to access knowledge and information. The impacts of using AI in agriculture span both developed and developing regions of the world, and have huge potential for improving food security, farmers incomes and their livelihoods globally.
In the wave of technological revolutions coming into the agricultural sector, we now have machines which target and spray weeds using near-infrared imagery, drones to gather huge volumes of data, satellites to predict drought patterns, and now AI mobile apps to diagnose crop diseases. Mobile phones are not just a communication tool, but a key agricultural tool of the future. Coupled with AI, this new technology can take on the role of an extension worker in regions where it may not otherwise be possible for advanced scientific knowledge and big-data to be just heard on the grapevine.