OliVaR: Improving olive variety recognition using deep neural networks

a Universidad de Cordoba, Cordoba, Spainb Sapienza Università di Roma, Dipartimento di Informatica, Rome, Italyc H.A.O. DEMETER, NAGREF, Institute of Olive Tree, Subtropical Crops and Viticulture, Chania, Crete, Greeced Council for Agricultural Research, Rende, Italye Olive Research Institute, Izmir, TurkeyA B S T R A C TThe easy and accurate identification of varieties is fundamental in agriculture, especially in the olive sector,where more than 1200 olive varieties are currently known worldwide. Varietal misidentification leads tomany potential problems for all actors in the sector: farmers and nursery workers may establish the incorrectvariety, leading to maladaptation in the field; olive oil and table olive producers may label and sell anonauthentic product; consumers may be misled; and breeders may commit errors during targeted crossingsbetween different varieties. To date, the standard for varietal identification and certification consists of twomethods: morphological classification and genetic analysis. The morphological classification consists of thevisual pairwise comparison of different organs of the olive tree, where the most important organ is consideredto be the endocarp. In contrast, different methods for genetic classification exist (RAPDs, SSR, and SNP).Both classification methods present advantages and disadvantages. Visual morphological classification requireshighly specialized personnel and is prone to human error. Genetic identification methods are more accuratebut incur a high cost and are difficult to implement.This paper introduces OliVaR, a novel approach to olive varietal identification. OliVaR used a teacher–student deep learning architecture to learn the defining characteristics of the endocarp of each specific olivevariety and perform varietal classification. We construct what is, to the best of our knowledge, the largest olivevariety dataset to date, comprising image data for 131 varieties from the Mediterranean basin. We thoroughlytest OliVaR on this dataset and show that it correctly predicts olive varieties with over 86% accuracy.