Computer Vision Accuracy Analysis with Deep Learning Model Using TensorFlow

Authors

  • T Tritva Jyothi Kiran Assistant Professor, Computer computer vision with creating a model that most accurately classifies the given image in the dataset and surpassing human performance. In previous research works many deep learning models are created and tested for image Classification on various datasets like MNIST, CIFAR-10, ImageNet using Python. Though they got good results of Accuracy for Classification, in this paper I have extended the work of measuring the performance analysis of Accuracy for Classification and also for the Predictions on CPU and GPU using TensorFlow2.0 and Keras on CIFAR-10 dataset having 50000 images of 10 datasets having a lot of different classes with very low resolutions. TensorFlow is an emerging technology on top of Python libraries developed by Google. This work reached an Accuracy 85% on GPU of Intel® Core™ i3- 7100U CPU which is acceptable with datasets used in this work are not easy to deal and all with very low resolutions having a lot of classes. That’s why it’s impacting the performance of the network. To classify and predict very low-resolution images from more datasets is really challenging one, it’s a great thing the computer vision accuracy performed excellent in my work. KEYWORDS- Accuracy, TensorFlow, GPU, Deep Learning, CNN, Classification, Prediction, Python, RELU, Backpropagation, Pooling, Flattening, Loss, Gradient Descent. I. INTRODUCTION The main challenge with such a large-scale image classification [13] task is that the diversity of the objects and classes. Thus, any model/algorithm that we use for this task must be able to handle very fine grained and specific classes, even though they may look very similar and are hard to distinguish.In more technical terms, we want to maximise the inter-class variability [1]. Image classification is a task that is associated with multi label assignments [1]. It involves the extraction of information from an image and then associating the extracted information to one or more class labels. Image classification within the machine learning domain can be approached as a supervised learning task. Manuscript received July 22, 2020 T. Tritva Jyothi Kiran, Assistant Professor, Computer Science Department, AKNU, Rajahmundry, Andhra Pradesh, India. Author

Keywords:

Accuracy, TensorFlow, GPU, Deep Learning, CNN, Classification, Prediction, Python, RELU, Backpropagation, Pooling, Flattening, Loss, Gradient Descent

Abstract

Deep learning has absolutely dominated  computer vision with creating a model that most  accurately classifies the given image in the dataset and  surpassing human performance. In previous research  works many deep learning models are created and tested  for image Classification on various datasets like MNIST,  CIFAR-10, ImageNet using Python. Though they got  good results of Accuracy for Classification, in this paper I  have extended the work of measuring the performance  analysis of Accuracy for Classification and also for the  Predictions on CPU and GPU using TensorFlow2.0 and  Keras on CIFAR-10 dataset having 50000 images of 10  datasets having a lot of different classes with very low  resolutions. TensorFlow is an emerging technology on  top of Python libraries developed by Google. This work  reached an Accuracy 85% on GPU of Intel® Core™ i3- 7100U CPU which is acceptable with datasets used in this  work are not easy to deal and all with very low  resolutions having a lot of classes. That’s why it’s  impacting the performance of the network. To classify  and predict very low-resolution images from more  datasets is really challenging one, it’s a great thing the  computer vision accuracy performed excellent in my  work. 

Downloads

Download data is not yet available.

References

Neha Sharma, Vibhor Jain, Anju Mishra, An Analysis of Convolutional Neural Networks for Image classification, Procedia Computer Science 132, 2018.

Mingyuan Xin1 and Yong Wang2, Research on image classification model based on deep convolution neural network, EURASIP Journal on Image and Technology and Applications, pp 587- 59.

Cristian Mateos jun-e Liu1 and Feng-Ping An, Image Classification Algorithm Based on Deep Learning Kernel Function, 2020.

D.lu & Q Weng, A survey of image classification methods and techniques for improving classification

Abu Sufian, Paramartha Dutta, Farhana sultana, Advancements in Image Classification using Convolutional Neural Network, 2018.

Md Tohidul islam B.M. Nafiz Karim Siddique Sagidur Rahman, Taskeed Jabid, Image Recognition with Deep Learning, 2018.

https://www.cs.toronto.edu/~kriz/cifar.html [9] Barry J. Wythoff, “Backpropagation neural networks: A tutorial”.

Ernest Istook, Tony Martinez, Improved backpropagation learning in neural networks with windowed momentum, International Journal of Neural Systems, vol. 12, no.3&4, pp. 303-318.

C. C. Jay Kuo, Understanding Convolutional Neural Networks with A Mathematical Model, University of Southern California, Los Angeles, CA 90089-2564, 2016.

Yann LeCun, LeonBottou, Yoshua Bengio, and Patrick Haffner, Gradient-Based Learning Applied to Document Recognition, PROC OF THE IEEE, NOVEMBER 1998.

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun, Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification, Microsoft Research, 2015.

Dominik Scherer, Andreas Muller, and Sven Behnke, Evaluation of Pooling Operations in Convolutional Architectures for Object Recognition, 20th International Conference on Artificial Neural Networks (ICANN), September 2010.

Jianxin Wu, Introduction to Convolutional Neural Networks, LAMDA Group, National Key Lab for Novel Software Technology, Nanjing University, 2017.

Y. lecu, B. boseter, J. SDenker, D Henderson, R.E. hovard, W. hubbred, AND L. Djackal, Backpropagation applied to handwritten zip code reconization, Neural Computation, VOL 1, NO.4PP.541-551, 1989.

B. H. Juang, Deep neural networks a developmental perspective, APSIPA Transactions on Signal and Information Processing, 2016.

F. Agostinelli, M. Hoffman, P. Sadowski, and P. Baldi, Learning activation functions to improve deep neural networks. arXiv:1412.6830, 2014.

Downloads

Published

2020-07-04

Issue

Vol. 8 No. 4 (2020): International Journal of Innovative Research in Computer Science and Technology (Jul) 2020

Section

Articles

How to Cite

Computer Vision Accuracy Analysis with Deep Learning Model Using TensorFlow . (2020). International Journal of Innovative Research in Computer Science & Technology, 8(4), 319–325. Retrieved from https://acspublisher.com/journals/index.php/ijircst/article/view/13243