A proposed 3-stage CNN classification model based on augmentation and denoising

Document Type : Research Paper


Department of Electrical Engineering, College of Engineering, University of Baghdad, Baghdad, Iraq


This work proposed a CNN classification model that aims to classify the faces by three stages applied to a real data set. The first stage shows the effects of the augmentation technique on the real data set where these effects include online, offline, and without augmentation. At this stage, the proposed CNN model is a built-from-scratch that has low computational complexity, low layers and the smallest filter sizes.  The second stage involved denoising the images in the real data set, where the images are preprocessed by applying the median, Gaussian, and mean filters to render the images more smooth and compare the effects of these filters based on the classification accuracy. The third stage involved a multi-class proposed model that contained 12 classes of images that were trained on the applied real data set, in addition to a benchmark set of images that was collected from the Internet. The findings reveal that the model accuracy reached $98.81\%$ when the offline augmentation model or the median filter was applied to the real data set, while it reached $97.48\%$ when the CNN multi-class proposed model was applied to identify the non-permission class. These processes were found to improve the performance parameters such as precision, recall, F1 score, and area under the curve (AUC).  Finally, to enhance the test prediction accuracy and test time, pre-training and fine-tuning (transfer learning) are applied on the real data set so as test accuracy and test time of our proposed model are better as compared with other models reached $99.7\%$ and 4 seconds respectively.


[1] S.H. Abdulredah and D.J. Kadhim, New approaches of cloud services access using Tonido cloud server for realtime applications, J. Engin. 26 (2020), no. 8, 83–99.
[2] B. Beddad and K. Hachemi, Efficient implementation of an improved median filter on TMS320C6416 digital signal processor, Proc. IEEE Int. Conf. Electric. Sci. Technol. Maghreb, Algiers, Algeria, 2018.
[3] L. Cuadros-Rodrigues, E. Perez-Castano and C. Ruiz-Samblas, Quality performance metrics in multivariate classification methods for qualitative analysis, TrAC Trends Anal. Chem. 80 (2016), 612–624.
[4] M. Hafiz Ishak, N.Sofia, M. Marzuki, M. Abdullah, Z. Soh, I. Isa and S. Sulaiman, Image quality assessment for image filtering algorithm: Qualitative and quantitative analyses, Proc. IEEE 9th Int. Conf. Cont. Syst. Comput. Engin., Penang, Malaysia, 2019, pp. 162–167.
[5] K. He, X. Zhang, S. Ren and J. Sun, Deep residual learning for image recognition, Proc. IEEE Conf. Comput. Vision Pattern Recog., 2016, pp. 770–778.
[6] J. Huber, Batch normalization in 3 levels of understanding, Towards Data Science, 2020.
[7] L. Hughes, M. Schmitt, L. Mou, Y. Wang, X. Zuh and R. Letters, Identifying corresponding patches in SAR and optical images with a pseudo-siamese CNN, IEEE Geosci. Remote Sens. Lett. 15 (2018), 784–788.
[8] S.Q. Jabbar, D.J. Kadhim, A proposed adaptive bitrate scheme based on bandwidth prediction algorithm for smoothly video streaming, J. Engin. 27 (2021), no. 1, 112–129.
[9] S.Q. Jabbar, D.J. Kadhim and Y. Li1, Developing a video buffer framework for video streaming in cellular networks, Wireless Commun. Mobile Comput. 2018 (2018).
[10] M.A. Joodi, M.H. Saleh and D.J. Kadhim, Increasing validation accuracy of a face mask detection by new deep learning model-based classification, Indones. J. Electric. Engin. Comput. Sci. 29 (2023), 304–3014.
[11] D.J. Kadhim and O.A. Hamad, Hamad Improving IoT applications using a proposed routing protocol, J. Engin. 20 (2014), no. 11, 50–62.
[12] N. Kan, N. Kondo, W. Chinsatit and T. Saitoh, Effectiveness of data augmentation for CNN-based pupil center point detection, Proc. IEEE 57th Ann. Conf. Soc. Instrum. Cont. Engin. Japan, Nara, Japan, 2018, pp. 441-464.
[13] D. Kingma and J. Ba, Adam: A method for stochastic optimization, 3rd Int. Conf. Learn. Represent., San Diego,2015.
[14] A. Krizhevsky, I. Sutskever and G.Hinton, Imagenet classification with deep convolutional neural networks, Commun. ACM 60 (2017), no. 6, 84–90.
[15] M. Kutlug¨un, Y. S¸irin and M. Karakaya, The effects of augmented training dataset on performance of convolutional neural networks in face recognition system, Proc. IEEE, Federated Conf. Comput. Sci. Inf. Syst., Leipzig, Germany, 2019, pp. 929—932.
[16] K. Lakhwani, H. Gianey and Sh. Gupta, An enhanced approach to improve UIQI and PSNR of noised colored images using DWTT filter, Proc. IEEE Int. Conf. Comput. Power Commun. Technol., Greater Noida, India, 2018, pp. 289–293.
[17] M. Ofori-Oduro and M. Amer, Data augmentation using artificial immune systems for noise-robust CNN models, Proc. IEEE Int. Conf. Image Process. (ICIP), Abu Dhabi, United Arab Emirates, 2020, pp. 833-837.
[18] J. Rasheed, E. Alimovski, A. Rasheed, Y. Sirin, A. Jamil and M. Yesiltepe, Effects of glow data augmentation on face recognition system based on deep learning, Proc. IEEE Int. Cong. Human-Comput. Interact. Optim. Robotic. Appl., Ankara, Turkey, 2020.
[19] S. Saxena, Introduction to Softmax for Neural Network, Analytic Vidhya, 2021.
[20] K. Simonyan and A. Zisserman, Very deep convolutional networks for large-scale image recognition, Comput. Vision Pattern Recog. Vers. 4, 6, Conference ICLR , 2015.
[21] C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens and Z. Wojna, Rethinking the inception architecture for computer vision, Proc. IEEE Conf. Comput. Vision Pattern Recog., 2016, pp. 2818–2826.
[22] A. Vikramathithan, S. Bhat and D. Shashikumar, Denoising high density impulse noise using Duo-Median filter for mammogram images, Proc. IEEE Int. Conf. Smart Technol. Comput. Electric. Electronic., Bengaluru, India, 2020, pp. 610–613.
[23] D. Villar, S. Torcida and G. Acosta1, Median filtering: A new insight, J. Meth. Imag. Vision 58 (2017), 130–146.
[24] D. Wang, D. Wang, Hongzhi Yu and G. Li, Face recognition system based on CNN, Proc. IEEE Int. Conf. Comput. Inf. Big Data Appl., Guiyang, China, 2020, pp. 470–473.
[25] Y. Weng and H. Zhou, Data augmentation computing model based on generative adversarial network, IEEE Access 7 (2019), 64223–64233.
[26] M. Wani, F. Bhat, S. Afzal and A. Khan, Advances in deep learning, Studies in Big Data, Springer, 2020.
[27] G. Wimmer, A. Uhl and A. Vecsei, Evaluation of domain specific data augmentation techniques for the classification of celiac disease using endoscopic imagery, Proc. IEEE 19th Int. Workshop on Multimedia Signal Proces., Luton, UK, 2017.
Volume 14, Issue 3
March 2023
Pages 121-140
  • Receive Date: 31 July 2022
  • Revise Date: 11 November 2022
  • Accept Date: 26 November 2022