Detect Plant Disease

Business Problem

The following excerpt (from the article https://arxiv.org/abs/1511.08060) summarizes this business problem: “human society needs to increase food production by an estimated 70% by 2050 to feed an expected population size that is predicted to be over 9 billion people. Currently, infectious diseases reduce the potential yield by an average of 40% with many farmers in the developing world experiencing yield losses as high as 100%.” The PlantVillage initiative from Penn State University aims at leveraging “the widespread distribution of smartphones among crop growers around the world with an expected 5 billion smartphones by 2020,” which “offers the potential of turning the smartphone into a valuable tool for diverse communities growing food.” The objective is to develop a mobile application to diagnose plant disease through machine learning and crowdsourcing, so that all smallholders can detect it early and act appropriately to help prevent the disease from spreading and impacting yield.

Intelligent Solution

Visual AI has the ability to create accurate models from example images that can then make predictions on new images. Similar to the way humans learn to identify patterns from visual cues, Visual AI will learn the patterns in images from plant leaves to accurately identify whether the plant is healthy or not and, if not healthy, determine the disease. This augments the role of people growing food and enables them to control yield losses. Advancements in model interpretability allow them to learn from the AI in return, as they can literally see what that AI is paying attention to when making its decisions. Contrary to traditional methods of computer vision, Visual AI does not require millions of labelled images. Both subject matter experts and data scientists can continuously iterate these models quickly by adding more images that capture more diseases over time.