Dataset of Avocado Leaf Images for Detection of Disease Caused by Thrips

Abstract

Agriculture is one of the few sectors that has yet to receive adequate attention from the auto-mathematical learning community. The importance of standard and publicly available datasets related to agriculture cannot be overemphasized and prevents practitioners in this discipline from taking full advantage of the benefits of these powerful computational prediction tools and techniques. To improve this situation, we developed, to the best of our knowledge, the first ready-to-use, off-the-shelf avocado data set available for future studies. The images come from avocado farmers' orchards in Quitasol-Abancay-Apurimac, one of the sixth largest avocado producing countries in the world. The dataset contains 3000 images of 1156 different leaves covering Trips disease and healthy avocado leaves. Although the dataset has been developed using only Quitasol Hass avocado leaves, given that this is the disease that affects the most in many countries, it is likely that this dataset can also be applied to identify avocado diseases in other countries, thus increasing avocado yields. It is expected that this dataset will attract the attention of researchers and practitioners of machine learning in the field of automated agriculture.