Due to the complex structure of the brain, detecting tumor areas on magnetic resonance images of the brain has always been an interesting topic. Therefore, various imaging techniques have been used to detect objects and with the recent advances in deep learning, the performance of object detection has been greatly improved. In this paper, a proposed convolutional neural network architecture model for classifying brain tumor types is proposed. Moreover, the performance of several existing object detection methods is evaluated. The proposed network structure was found to deliver significant performance with an overall best accuracy of 96.05%. Therefore, the results indicate the ability of the proposed model to classify brain tumors for several purposes, moreover, these results confirm that appropriate preprocessing and data augmentation will lead to an accurate classification.