With the developing technology, the concepts of "Artificial Intelligence" and "Deep Learning", which we have been hearing fre quently in recent years, have many application areas. These methods, which imitate human intelligence, learn the data sets they receive from their environment through experience, just like humans. In this study, apple varieties grown in Yahyalı district of Kayseri were classified according to their type variables. Since the apple fruit is picked in September, it was obtained at the harvest time when the apples were plucked from the tree. The biggest problem of the farmers producing apples is the classification of apples according to their varieties without a handprint and as soon as possible. In this study, a total of 120 images were taken from 20 Gold en, 20 Argentina, 20 Buckeye Gala, 20 Galaval, 20 Superchief and 20 Joremin apple varieties. The images were taken with a Canon EOS 70D DSLR camera at the same angle and the same size, on a fixed background. R2021a version of MATLAB program was used for image processing. Deep learning algorithms were used to classify apple varieties according to their types. AlexNet and GoogleNet deep learning algorithms, which are among the most basic architectures used in solving classification problems, are used. The study was carried out in both AlexNet and GoogleNet methods at 10 epochs and sgdm training algorithm. Learning rates are taken as 0.0001 and 0.0003 for Ale xNet and GoogleNet, respectively. 70% of the images were used for training and 30% for testing, and the total data set consists of 120 images, 20 of each type. AlexNet architecture has 83.33% success rate, 1 minute 52 seconds. The classification success rat e of the GoogleNet architecture is 91.67%, and it performed the most successful classification process for 2 minutes and 14 seconds.