Özet:

The classification of plant species from the satellite images provides very important information for the management of agricultural areas, food security and the creation of agricultural policies. Plant species are generally estimated on the basis of vegetation index calculated from satellite images or using various image classification techniques. But the success rate in these approaches is low because different plants have similar phenological and spectral properties. Therefore, a new, more accurate and more successful approach is needed in the process of classification of plant species by satellite images. The aim of this study is to classify plant species from satellite images using the Rassal Forest (RO), Support Vector Machine (DVM) and K-NN (K-N) machine learning algorithms. In the framework of the study, Sentinel 2 satellite images in the time series form were used in the classification of plants grown in the agricultural areas of the 2020 Gökhöyük Agricultural Management Directorate. According to the results obtained from the study, the most successful classification (%95.3) was achieved with DVM (%75.9) while the lowest success was calculated with RO. The success of classification with K-NN is 91.8%.

Anahtar Kelimeler: