Mobile Application Development for Optimal and Rapid Diagnosis of Vine Diseases

In a perennial culture such as the vine, late and inaccurate identification of the damage created by diseases and pests can cause significant crop loss or even the loss of the entire production. To support the growers in Romania, in 2010, we printed and also posted online on www.scvblaj.ro a guide on how to diagnose vine diseases and pests that describes in detail the most important pathogens together with pictures and various solutions in order to help the growers to bring the plants back to a healthy state whenever possible. Given the high number of growers in our country, especially small family growers, the guide is addressed to anyone, regardless of their past experience. Between 2010 and 2014, thousands of growers have accessed the online version of the guide and contacted us to request additional help. We gathered a lot of valuable information during that time but most notably the fact that many growers were contacting us too late, when a disease had already spread beyond any hope of recovery or worse after a wrong treatment was applied because of an inaccurate diagnosis. In recent years we noticed an increasing number of wrong diagnostics of vine diseases and a tendency for pathogens to be more aggressive. Thus finding a new solution, easily accessible to growers that would offer a faster, easier and up to date diagnosis, was necessary. Recent developments in the field of machine learning (Ghahramani, 2015), breakthroughs in computer vision (Krizhevsky et al., 2012) and the availability of cheap computational hardware has led us to look for a way to automatically detect and classify a disease just by having a computer or a mobile phone process a picture of an affected plant. Given the widespread availability of mobile devices, we believe we can help the vine growers further.