Crop Disease Images Collected, Available Via Smartphones

PlantVillage images show a healthy tomato leaf, upper left, next to leaves with symptoms of disease. (David Hughes, Pennsylvania State University)

25 November 2015. A team in the U.S. and Switzerland is assembling a database of 50,000 images of plant diseases for a smartphone app to help farmers worldwide deal with those diseases. Entomologist David Hughes at Pennsylvania State University in University Park and epidemiologist Marcel Salathé at Ecole polytechnique fédérale de Lausanne or EPFL, the Swiss Federal Institute of Technology in Lausanne, are leading the team creating the open-access database, known as PlantVillage.

Hughes, Salathé, and colleagues are developing PlantVillage as a tool for growers to take advantage of their smartphones to capture and share images of crop disease in their fields, as well as learn new techniques for battling those diseases. PlantVillage is aimed particularly at small subsistence farmers with limited resources for investing in new technology, but who can use their smartphones to get answers about diseased crops. The U.N.’s Food and Agricultural Organization estimates that plant diseases and pests reduce annual crop yields worldwide by 20 to 40 percent.

Collecting the images is a first step in developing PlantVillage’s utility for growers. A key feature of PlantVillage is machine learning algorithms that detect patterns in the collected images and data, and provide responses to queries. The system’s developers envision farmers using their smartphones to snap and upload a photo of a disease plant, which the PlantVillage database and algorithms would match, analyze, and return with guidance via the app, which is now available for iPhones.

Hughes and Salathé say PlantVillage is one of the largest free libraries on plant diseases, covering 154 crops and more than 1,800 diseases. “In addition to being a library, PlantVillage is a network of experts who help people around the world find solutions to their problems,” says Salathé in a joint statement. “Our goal is to let the smartphone do most of the diagnosis, so that human experts can focus on the unusual and difficult cases.”

The algorithms remains a bottleneck in PlantVillage’s development, due to the relatively small number of images collected so far to train algorithms to discriminate between diseased and healthy crops. But the developers plan to use the project’s open-source and open-access properties to solve that problem. “By providing all these images with open access, we are challenging the global community in two ways,” notes Hughes. “We are encouraging the crop-health community to share their images of diseased plants, and we are encouraging the machine learning community to help develop accurate algorithms.”

The developers plan to hold a crowdsourced online competition to create more algorithms that diagnose crop diseases. Hughes tells more about PlantVillage in the following video.

• Smartphone Add-On Found Comparable to Pro Stethoscopes
• Mobile App Collecting Placenta Health Data
• Inside Air Monitor Devised as Smartphone Add-On
• University Faculty Design Mood-Tracking Mobile App
• Smartphone Vision App Found Accurate as Clinic Test

*     *     *