A study on effects of Data Augmentation in Classification Accuracy in Deep Learning.

Abstract

The data augmentation is an important part of deep learning and computer vision. The techniques of data augmentation are also a widely used method in the most popular deep learning models. However there are very little studies done on the effects of data augmentation techniques have on the accuracy of the model. This thesis is mostly based on investigational study on data augmentation in classification accuracy in deep learning. The objective of this thesis is to investigate the different data augmentation techniques and their effects in learning process of VGGNet model, performance generalization, prediction and other behaviour of deep neural networks.