MIDS Capstone Project Spring 2021

Plant Disease Detection with Unsupervised Learning

Medicrop is focused on lack of easy access to plant disease expertise leading to crop loss, which accounts for a global 15% reduction in crop yield and disproportionately impacts developing regions. Using publicly available images of plant disease combined with ML-generated supplemental images, we apply a combination of unsupervised learning and supervised learning for our MVP that we intend to launch globally in a free beta-testing format by April 2021.

"The constant attack of pest and diseases represents a challenge for food production and food security in LATAM. The rapid detection of those phytosanitary problems are key for farmers to be able to apply a timely control and avoid the spread of the problem. At the field level farmers, extension officers, agricultural advisors normally faced the challenges of identifying properly the major problems in order to be able to recommend the best management practices. However, when the problem can´t be identified at the field using their knowledge, they normally use the local-national diagnostic laboratories. Depending on the country the results can be obtained in several days or weeks. Having a Diagnostic App, as the one you have showed me, which can facilitate/accelerate the diagnostic process with a good % of confidence/reliability will be very well receive by farmers and extension officers! Certainly, it will be a valuable tool to have!" - Dr. Colmenarez; CABI Brazil Centre Director & Plantwise Regional Coordinator – Latin America and Caribbean

Last updated:

April 14, 2021