Turkey provides approximately 75% of world hazelnut production. Therefore, accurate determination of hazelnut areas is important for the amount of hazelnut production and productivity studies. In this study, it was aimed to determine the hazelnut orchards using the Sentinel-2 satellite image dated 02.07.2019. In this context, Besikduzu District of Trabzon province was selected as the study area and district center as the application area. Pixel-based image classification method is used for land detection. Three different machine learning algorithms random forest (RF), K nearest neighbor (KNN) and support vector machines (SVM) were applied to identify hazelnut gardens. General accuracy values for classifications were determined as 85,3870, 87,5398 and 91,0948 for RF, KNN and SVM algorithms, respectively.