Research Article
BibTex RIS Cite

Görüntü Sınıflandırmasında Bulanıklığın Yan Etkilerinin Evrişimsel Sinir Ağları ile Analizi

Year 2025, Volume: 3 Issue: 1, 32 - 37, 30.06.2025

Abstract

Bulanıklık, görüntü kalitesini bozan yaygın faktörlerden biridir ve hareket, odak kaybı veya çevresel koşullar gibi çeşitli etkenlerden kaynaklanabilir. Bir veri kümesinde kısmen veya tamamen bulanık görüntülerin bulunması, nesne tanımayı zorlaştırarak görüntü sınıflandırma modellerinin etkinliğini azaltabilir. Bu sorunu hafifletmek için bulanık görüntüler ya veri kümesinden çıkarılmalı ya da görüntü netleştirme (deblurring) teknikleriyle işlenmelidir. Bu projede, bulanık görüntülerin derin öğrenme tabanlı görüntü sınıflandırma modellerinin performansı üzerindeki etkisi incelenmiştir. Özellikle, farklı bulanıklık seviyelerinin sınıflandırma doğruluğunu nasıl etkilediği analiz edilmiştir. Bu amaçla, CIFAR-10 veri kümesi kullanılarak bulanıklık oranları %0, %25, %50 ve %100 olan veri kümeleri ile bir evrişimli sinir ağı (CNN) modeli eğitilmiştir. Deney sonuçları, eğitim veri kümesindeki bulanık görüntü oranı arttıkça doğrulama (validation) doğruluğunun azaldığını göstermiştir. Model, %0, %25, %50 ve %100 oranlarında bulanık görüntü içeren veri kümeleriyle eğitildiğinde sırasıyla %67.53, %65.50, %63.90 ve %55.74 doğrulama doğrulukları elde etmiştir. Bu bulgular, görüntü bulanıklığının sınıflandırma performansı üzerindeki olumsuz etkilerini ortaya koymakta ve derin öğrenme uygulamalarında yüksek kaliteli görüntü verisinin önemini vurgulamaktadır.

References

  • [1] Su B, Lu S, Tan CL. Blurred image region detection and classification. In: Proceedings of the 19th ACM International Conference on Multimedia; 2011. New York: Association for Computing Machinery. p. 1397–1400.
  • [2] Liu R, Li Z, Jia J. Image partial blur detection and classification. In: 2008 IEEE Conference on Computer Vision and Pattern Recognition; 2008 Jun; Anchorage, AK. p. 1–8.
  • [3] Gowthami S, Harikumar R. Conventional neural network for blind image blur correction using latent semantics. Soft Comput 2020;1–15.
  • [4] De Vylder J, et al. Image restoration using deep learning. In: Annual Machine Learning Conference of Belgium and The Netherlands (Benelearn 2016); 2016.
  • [5] Wang H, Yue Z, Zhao Q, Meng D. A deep variational Bayesian framework for blind image deblurring. IEEE Trans Pattern Anal Mach Intell 2021;43(6):1971–1984.
  • [6] Zhang K, Ren W, Luo W, Lai WS, Stenger B, Yang MH, Li H. Deep image deblurring: A survey. IEEE Trans Pattern Anal Mach Intell 2023;45(1):1–20.
  • [7] Ren D, Zhang K, Wang Q, Hu Q, Zuo W. Neural blind deconvolution using deep priors. IEEE Trans Image Process 2020;29:4379–4391.
  • [8] Yoneyama A, Minamoto T. No-reference image blur assessment in the DWT domain and blurred image classification. In: 2015 12th Int Conf on Information Technology – New Generations; 2015 Apr; Las Vegas, NV. p. 329–334.
  • [9] Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Commun ACM 2017;60(6):84–90.
  • [10] Krizhevsky A, Hinton GE. Learning multiple layers of features from tiny images. Tech Rep, University of Toronto; 2009.
  • [11] Cheng J, Chen WT, Lu X, Yang MH. BD-Diff: Generative diffusion model for image deblurring on unknown domains with blur-decoupled learning. arXiv preprint arXiv:2502.01522; 2025.
  • [12] Amanturdieva A, Zafar MH. Efficient transformer for high-resolution image motion deblurring. arXiv preprint arXiv:2501.18403; 2025.
  • [13] Xiang Y, Zhou H, Li C, Sun F, Li Z, Xie Y. Deep learning in motion deblurring: current status, benchmarks and future prospects. arXiv preprint arXiv:2401.05055; 2024.

Analyzing the Side Effects of Blur in Image Classification with Convolutional Neural Networks

Year 2025, Volume: 3 Issue: 1, 32 - 37, 30.06.2025

Abstract

Blur is one of the common factors that deteriorate image quality and can be caused by various factors such as motion, defocus, or environmental conditions. The presence of partially or globally blurred images in a dataset can make object recognition challenging, thereby reducing the effectiveness of image classification models. To mitigate this issue, blurred images must either be removed from the dataset or processed using deblurring techniques. In this project, the impact of blurred images on the performance of deep learning-based image classification models investigated. Specifically, the goal was to analyze how different levels of image blur affect classification accuracy. To achieve this, a convolutional neural network (CNN) model was trained using the CIFAR-10 dataset, with varying proportions of blurred images: 0%, 25%, 50%, and 100%. The experiment results demonstrated that increasing the proportion of blurred images in the training dataset led to a decline in validation accuracy. The model achieved validation accuracies of 67.53%, 65.50%, 63.90%, and 55.74% when trained with datasets containing 0%, 25%, 50%, and 100% blurred images, respectively. These findings highlight the adverse effects of image blur on classification performance, emphasizing the importance of high-quality image data in deep learning applications.

References

  • [1] Su B, Lu S, Tan CL. Blurred image region detection and classification. In: Proceedings of the 19th ACM International Conference on Multimedia; 2011. New York: Association for Computing Machinery. p. 1397–1400.
  • [2] Liu R, Li Z, Jia J. Image partial blur detection and classification. In: 2008 IEEE Conference on Computer Vision and Pattern Recognition; 2008 Jun; Anchorage, AK. p. 1–8.
  • [3] Gowthami S, Harikumar R. Conventional neural network for blind image blur correction using latent semantics. Soft Comput 2020;1–15.
  • [4] De Vylder J, et al. Image restoration using deep learning. In: Annual Machine Learning Conference of Belgium and The Netherlands (Benelearn 2016); 2016.
  • [5] Wang H, Yue Z, Zhao Q, Meng D. A deep variational Bayesian framework for blind image deblurring. IEEE Trans Pattern Anal Mach Intell 2021;43(6):1971–1984.
  • [6] Zhang K, Ren W, Luo W, Lai WS, Stenger B, Yang MH, Li H. Deep image deblurring: A survey. IEEE Trans Pattern Anal Mach Intell 2023;45(1):1–20.
  • [7] Ren D, Zhang K, Wang Q, Hu Q, Zuo W. Neural blind deconvolution using deep priors. IEEE Trans Image Process 2020;29:4379–4391.
  • [8] Yoneyama A, Minamoto T. No-reference image blur assessment in the DWT domain and blurred image classification. In: 2015 12th Int Conf on Information Technology – New Generations; 2015 Apr; Las Vegas, NV. p. 329–334.
  • [9] Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Commun ACM 2017;60(6):84–90.
  • [10] Krizhevsky A, Hinton GE. Learning multiple layers of features from tiny images. Tech Rep, University of Toronto; 2009.
  • [11] Cheng J, Chen WT, Lu X, Yang MH. BD-Diff: Generative diffusion model for image deblurring on unknown domains with blur-decoupled learning. arXiv preprint arXiv:2502.01522; 2025.
  • [12] Amanturdieva A, Zafar MH. Efficient transformer for high-resolution image motion deblurring. arXiv preprint arXiv:2501.18403; 2025.
  • [13] Xiang Y, Zhou H, Li C, Sun F, Li Z, Xie Y. Deep learning in motion deblurring: current status, benchmarks and future prospects. arXiv preprint arXiv:2401.05055; 2024.
There are 13 citations in total.

Details

Primary Language English
Subjects Data Engineering and Data Science
Journal Section Research Articles
Authors

Ahmet Fırat Yelkuvan 0000-0003-4148-1923

Publication Date June 30, 2025
Submission Date April 7, 2025
Acceptance Date May 5, 2025
Published in Issue Year 2025 Volume: 3 Issue: 1

Cite

IEEE A. F. Yelkuvan, “Analyzing the Side Effects of Blur in Image Classification with Convolutional Neural Networks”, Sivas Cumhuriyet Üniversitesi Mühendislik Fakültesi Dergisi, vol. 3, no. 1, pp. 32–37, 2025.