Measuring the Effect of Data Augmentation in a CNN-Based Deep Neural Network Model

Abstract

Traditional classification methods have difficulty in meeting the changing needs according to the ever-increasing data piles. With the development of processors with high performance as memory and processing capabilities, deep learning-based methods have been widely used. A large amount of data is needed to train a deep learning-based model, which is a computational science field. CIFAR-10, which contains images of 10 different objects in the world, is a benchmark dataset used effectively in image identification and classification. The proposed deep learning-based models should be tested in a computer environment in order to be used in real life. The proposed model performs the testing process with images that it has never encountered during the training phase. In this article, a deep learning model is proposed that performs classification on the CIFAR-10 dataset, which contains images of objects in the world. An effective classification method has been developed by removing the overfitting effect, if any, on the proposed model. Proposed model, classification process was carried out both with and without data augmentation. The data set used was expanded with random crop, scale transformation, vertical and horizontal flipping data augmentation techniques. In the experimental studies, there was a big difference between the performance of the process using the data augmentation technique and the process without any augmentation. Using different augmentation techniques together or individually did not improve model performance. Proposed model achieved success rates of 91.93%, 93.63% and 90.49%, respectively, including train accuracy, precision, recall. According to the results obtained, it can be said that the study has achieved results that can compete with the literature.