Citation: Petrellis, N.; Antonopoulos, C.; Keramidas, G.; Voros, N. Mobile Plant Disease Classifier, Trained with a Small Number of Images by the End User. Agronomy 2022, 12, 1732. https://doi.org/10.3390/ agronomy12081732 Academic Editors: Zhanyou Xu, Reka Howard and Lizhi Wang Received: 15 June 2022 Accepted: 18 July 2022 Published: 22 July 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). agronomy Article Mobile Plant Disease Classifier, Trained with a Small Number of Images by the End User Nikos Petrellis 1, * , Christos Antonopoulos 1 , Georgios Keramidas 2 and Nikolaos Voros 1 1 Embedded System Design & Automations Lab (ESDALAB), Electrical and Computer Engineering Department, University of Peloponnese, 26334 Patras, Greece; ch.antonop@esdalab.ece.uop.gr (C.A.); voros@esdalab.ece.uop.gr (N.V.) 2 School of Informatics, Aristotle University of Thessaloniki, 54642 Thessaloniki, Greece; gkeramidas@csd.auth.gr * Correspondence: npetrellis@uop.gr Abstract: Mobile applications that can be used for the training and classification of plant diseases are described in this paper. Professional agronomists can select the species and their diseases that are supported by the developed tool and follow an automatic training procedure using a small number of indicative photographs. The employed classification method is based on features that represent distinct aspects of the sick plant such as, for example, the color level distribution in the regions of interest. These features are extracted from photographs that display a plant part such as a leaf or a fruit. Multiple reference ranges are determined for each feature during training. When a new photograph is analyzed, its feature values are compared with the reference ranges, and different grades are assigned depending on whether a feature value falls within a range or not. The new photograph is classified as the disease with the highest grade. Ten tomato diseases are used as a case study, and the applications are trained with 40–100 segmented and normalized photographs for each disease. An accuracy between 93.4% and 96.1% is experimentally measured in this case. An additional dataset of pear disease photographs that are not segmented or normalized is also tested with an average accuracy of 95%. Keywords: disease classification; training; mobile application; image processing; segmentation; tomato diseases 1. Introduction The economic and ecological overhead of crop management can be significantly in- creased if plant diseases are not detected and treated early. Larger amounts of pesticides are required if a disease has already spread, and part of the production may be perma- nently lost. The plants need to be continuously monitored in order for the initial disease symptoms to be recognized. Medium and small producers may not be able to afford to hire professional agronomists for long periods. Several precision agriculture and Internet of Things (IoT) tools have recently been proposed that can assist producers in efficiently mon- itoring the health of their plants. Although most of these tools cannot provide a certified diagnosis, they can guide the producer to perform additional, more reliable tests [1], if it is necessary, to verify a disease infection. The diagnosis of a plant disease can be based on its symptoms: under- or over-development of tissues or necrosis or alternation in the appearance of plant parts. The progression of the primary or secondary symptoms depends on biotic agents. Similar symptoms may also be caused by the use of herbicides, chemicals and air pollution. Some approaches require advanced tools and hardware or a combination of infor- mation derived from multiple sources (e.g., sensors) to perform a reliable plant disease diagnosis. Image processing can be combined with sensor indications, meteorological data and information entered by the producers in response to questionnaires. Unmanned Air Agronomy 2022, 12, 1732. https://doi.org/10.3390/agronomy12081732 https://www.mdpi.com/journal/agronomy