Leveraging Rapid Advances in Computer Science to Feed the World and Preserve the Environment

A person taking a photo of a tomato plant leaf with a smartphone
Project Type
April 2016
A team of researchers at Penn State is using machine learning to develop an app that can diagnose plant diseases using images taken on a smartphone.

The goal of this project is to enable the development of machine learning algorithms to digitally diagnose plant diseases and position Penn State as a leader in automated diagnosis of plant diseases via phones. This project builds upon the largest collection of curated images of diseased and healthy plants (currently at 85,000 and growing) and the largest social network around crop health (www.PlantVillage.org, currently over 2 million users and growing at 250% per year). In this project three diverse researchers (Hughes, Wang, and Giles) will work with Google and other computer scientists to improve upon image recognition algorithms by using additional information contained within plant pathology publications. Google has already developed an algorithm based on 18,000 images that Hughes provided and this has an accuracy of 93%. Part of this project is testing this algorithm under different conditions and building upon it with more images and literature derived data (publications on plant pathology that provide additional knowledge of disease occurrence by location). 

We want to use our seed grant funding to gather more data to pitch proposals to both federal agencies (NSF/USDA) as well as donors such as the Gates Foundation, Rockefeller Foundation, Buffett Foundation, and national aid bodies (USAID, Irish Aid, DIFID). This latter effort (foundations) is to be coordinated through the Office of the Vice President for Research at Penn State and in collaboration with a former Penn State plant pathologist who now heads up an eight-country West African Network to reduce disease in staple crops (currently funded by the Gates Foundation). 


James Wang

Professor of Information Sciences and Technology, College of Information Sciences and Technology