Image Classification using Convolution Neural Network
Tapan Bhavsar1, Bhavinkumar Gajjar2
1Tapan Bhavsar, Student, Department of Electronics and Communication Engineering, Indus Institute of Technology and Engineering, Indus University, Ahmadabad (Gujarat), India.
2Bhavinkumar Gajjar, Assistant Professor, Department of Electronics and Communication Engineering, Indus Institute of Technology and Engineering, Indus University, Ahmadabad (Gujarat), India.
Manuscript received on 13 June 2017 | Revised Manuscript received on 20 June 2017 | Manuscript Published on 30 June 2017 | PP: 192-195 | Volume-6 Issue-5, June 2017 | Retrieval Number: E5038066517/17©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: Convolution neural network has been mostly used for image classification in machine learning and computer vision. In simple neural network, single layer’s feature may not contain enough useful information to predict image class correctly [6]. Using a feed forward CNN, misclassification rate can be reduced by some additional layers that contain acceptable information to predict image class. Also gradient based learning algorithm can be improved to synthesize complex decision that classify high dimensional pattern such as object edges and shape. In this paper, we make effort to modify standard neural network to transfer more information layer to layer. Moreover, already learned CNN model with training images are used to extract features from multiple layers. In this experiment, MNIST and CIFAR 10 dataset have been used to classify random images in 10 different classes labelled airplane, automobile, bird, cat, deer, dog, frog, horse, ship and truck. In the addition, GPU can train CNN faster without giving the preference to hardware.
Keywords: Convolution Neural Network, CIFAR 10, gradient Based Learning Algorithm, Image Classification, MNIST, Machine Learning
Scope of the Article: Classification