Deep Learning and Transfer Learning Approaches for Image Classification
Sajja Tulasi Krishna1, Hemantha Kumar Kalluri2
1Sajja Tulasi Krishna, Vignan’s Foundation for Science Technology and Research, Deemed to be University, India.
2Hemantha Kumar Kalluri, Vignan’s Foundation for Science Technology and Research, Deemed to be University, India.
Manuscript received on 14 February 2019 | Revised Manuscript received on 05 March 2019 | Manuscript Published on 08 June 2019 | PP: 427-432 | Volume-7 Issue-5S4, February 2019 | Retrieval Number: E10900275S419/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Deep Learning is-one of the machine learning areas, applied in recent areas. Various techniques have been proposed depends on varieties of learning, including un-supervised, semi-supervised, and supervised-learning. Some of the experimental results proved that the deep learning systems are performed well compared to conventional machine learning systems in image processing, computer vision and pattern recognition. This paper provides a brief survey, beginning with Deep Neural Network (DNN) in Deep Learning area. The survey moves on-the Convolutional Neural Network (CNN) and its architectures, such as Le Net, Alex Net, Google Net, VGG16, VGG19, Resnet50 etc. We have included transfer learning by using the CNN’s pre-trained architectures. These architectures are tested with large ImageNet data sets. The deep learning techniques are analyzed with the help of most popular data sets, which are freely available in web. Based on this survey, conclude the performance of the system depends on the GPU system, more number of images per class, epochs, mini batch size.
Keywords: Convolutional- Neural Network (CNN); Deep-Learning (DL);-Machine Learning (ML); Pre-trained Network; Transfer Learning.
Scope of the Article: Deep Learning