Enhanced YOLOv11n with Ghost Convolution for Automatic Detection of Pistachio Leaf Disease

According to public data, Turkey is the thirdlargest producer of the pistachio, followed by the USA and Iran in the last ten years. Especially in Gaziantep province, the production of pistachio takes a significant role in terms of national cuisine, exporting, and employment, in Turkey. On the other hand, nowadays technological advancements in artificial intelligence put forward a number of conveniences in agricultural areas. In this context, automatic and early detection of plant diseases is one of the most beneficial properties of robust object detection models. In this study, a new dataset is created for detecting pistachio leaf disease called forda formicaria or gal afidi (gall aphid) with improved YOLOv11n architecture, which is obtained with the help of added Ghost Convolution modules. Moreover, this developed model is compared with state-of-theart object detection models such as YOLOv5n, YOLOv10n, YOLOv11n, and YOLOv12n with respect to precision, recall, F1-score, [email protected], and [email protected]:95. When the results are examined, it is clearly seen that the proposed model gives higher performance results for most of the calculated test values.