Enhancing Performance of Deep Learning Models with different Data Augmentation Techniques: A Survey
Cherry Khosla, Baljit Singh Saini
Abstract
Deep convolutional neural networks have shown impressive results on different computer vision tasks. Nowadays machines are fed by new artificial intelligence techniques which makes them intelligent enough to cognize the visual world better than humans. These computer vision algorithms rely heavily on large data sets. Having a large training data set plays a very crucial role in the performance of deep convolutional neural networks. We can enhance the performance of the model by augmenting the data of the image. Data augmentation is a set of techniques that are used to increase the size and quality of the image with label preserving transformations. This survey paper focuses on different data augmentation techniques based on data warping and oversampling. In addition to data augmentation techniques, this paper gives a brief discussion on different solutions of reducing the overfitting problem.