A.I. is here to save your salad.
If a farmer finds orange pustules on a crop of corn, how long does it take to determine whether it’s southern rust or another disease? That reaction time has big implications for the world’s food supply.
Every year, 20 to 40 percent of global crop yields are lost due to pests and diseases, according to the United Nations. It makes it harder to fight these afflictions when farmers have to spend time identifying what exactly is afflicting their crops. This delays them from taking the necessary steps to remedy the situation, affecting their crop output and, hence, the food supply. That’s where artificial intelligence can help.
Farmwave, an app using artificial intelligence that was released in beta in July, allows a farmer in the field to take a photo of a suspect crop and receive a diagnosis in seconds rather than days. Armed with this information, farmers can combat diseases and pests quicker, addressing the issue before it spreads and saving more money for the farmer and more crops from the trash. “Having greater yields helps farmers, especially small landholders, really preserve their crops,” says Chris Chan, chief operations officer at Farmwave. “That means more people in their community can eat.”
Farmwave, developed by the Atlanta-based technology company CAMP3, has partnered with the agricultural program at the University of Georgia (UGA) to use its thousand-strong photo bank of crops with different diseases and pests. The UGA has collected these photos from farmers throughout the state since the 1990s.
Mark McCann, assistant dean for Extension at the UGA’s College of Agricultural & Environmental Sciences, says Farmwave will allow farmers to self-diagnose in the field. “If it’s a plant disease or rapid pest, the quicker we can get information, the better, particularly in produce crops, where diseases can move through a high-value crop very quickly,” he says. “It’s quite expensive, so a fast response can reduce the impact of those diseases in the field.”
Making Farmwave work involves teaching it to recognize diseases and pests by feeding it one photo after another, all from different angles and during different times of the day, says Chan. In addition to using the UGA’s catalogue, Farmwave is working on agreements with other universities to take more current photos of crops. After collecting the images and training the app based on those photos, each crop, disease and pest must go through testing. In addition to Farmwave’s in-house team, the main testing takes place in Iraq, Jordan and France.
Farmwave’s first crop is corn because of its wide growth, followed by soybeans, says Chan. Right now, Farmwave can identify southern rust on corn. The 12 remaining corn diseases and pests will likely be deployed to the app by the end of 2018, along with some soybean afflictions. Farmwave is also working on wheat, canola, tomato and a slew of other crops.
The ability to quickly identify crop diseases and pests speeds up the process that farmers now follow. Take Georgia’s rural Madison County, for example, which grows crops like hay, soybeans and wheat. Adam Speir serves as the UGA’s Extension coordinator and agriculture agent for the county, helping local farmers with any issues that may arise. If a farmer needs to identify a disease, he contacts Speir.
Nowadays, that’s usually by text. A farmer takes a photo of a blighted crop and sends it to Speir for identification. Sometimes Speir can recognize the issue immediately; other times, he forwards the photo to a UGA expert. That process can take one to two days or longer, occasionally requiring Speir to make a farm visit.
But a tool like Farmwave? Though it can’t always replace a person, it can cut the time involved, says Speir. “Anything that can help a farmer identify an issue as quickly as possible is beneficial,” he says. “In farming, time is money. The longer it takes for a farmer to identify and correct an issue — whether that’s finding a certain fungicide or insecticide or determining a management practice — the more it can potentially cost that producer in yield or the stand of the crop growing successfully and being profitable to that farmer.”
The U.S. system of extension agents — university-employed agriculture experts — exists in every state, but not all are as robust as those found in Georgia. Some countries have no resources at all — a knowledge gap that Farmwave hopes to fill. Chan says the app’s next phase is a communication feature between users, which would allow, say, a Jordan Valley farmer to trade lettuce tips with a Wisconsin farmer.
McCann says Farmwave is only in the beginning stages of what it can do. One day, it will be capable of not only identifying diseases but also teaching farmers along the way. “If they learn that this is what a particular disease looks like, well, the next time, they’ll know,” he says. “It’s a diagnosis tool in the field, but it’s also training.”