International Journal of Image Processing and Pattern Recognition Original Research

Toward Real-Time Traffic Sign Recognition: An Efficient CNN Approach

  1. Samkit Jain Student, Maharaja Surajmal Institute, New Delhi
  2. Harjender Singh ✉ Assistant Professor, Maharaja Surajmal Institute, New Delhi

Abstract

Ensuring that vehicles correctly interpret traffic signage is essential for safe autonomous navigation, as incorrect or delayed recognition can directly affect real-time driving decisions. However, correctly identifying traffic signs from images is not straightforward. Variations in lighting conditions, viewing angles, object size, partial occlusions, and complex backgrounds make the classification task considerably challenging. This work introduces a convolution-based deep learning frame-work designed for multi-class traffic sign recognition. The pro-posed model is validated on the German Traffic Sign Recognition Benchmark (GTSRB), a publicly available dataset containing 43 labeled traffic sign classes.. Before training, all images are resized to 50×50 pixels to maintain consistent input dimensions, and pixel values are scaled to the range [0, 1] to improve numerical stability and learning efficiency. The network is structured to extract hierarchical visual patterns through successive convolutional layers equipped with ReLU activation. To progressively condense spatial information and emphasize dominant features, max-pooling operations are in-corporated between convolutional stages. Dropout regularization is applied during training to improve generalization and limit overfitting. Following feature extraction, the resulting representations are reshaped into a one-dimensional vector and forwarded to fully connected layers responsible for high-level decision making. Class probabilities are generated using a Softmax function at the output stage. Model training is carried out using the Adam optimiDuring training, the network attained an accuracy of 93.39%, while validation performance reached 99.12*%, reflect-ing strong generalization on unseen data. When evaluated on a separate test dataset, the system obtains an overall accuracy of 94.94%, along with macro-averaged precision, recall, and F1-scores of 94.41%, 93.42%, and 93.60%, respectively. These results suggest that the developed CNN model provides reliable and consistent performance, making it suitable for practical traffic sign recognition scenarios.

Keywords

References (18)

  1. J. Stallkamp, M. Schlipsing, J. Salmen, and C. Igel, “Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition,” Neural Networks, vol. 32, pp. 323–332, 2012.
  2. Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proc. IEEE, vol. 86, no. 11, pp. 2278–2324, Nov. 1998.
  3. K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” in Proc. ICLR, 2015.
  4. K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. CVPR, 2016, pp. 770–778.
  5. M. Tan and Q. Le, “EfficientNet: Rethinking model scaling for convo-lutional neural networks,” in Proc. ICML, 2019, pp. 6105–6114.
  6. A. Sermanet and Y. LeCun, “Traffic sign recognition with multi-scale convolutional networks,” in Proc. IJCNN, 2011, pp. 2809–2813.
  7. A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv:2004.10934, 2020.
  8. S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards real-time object detection with region proposal networks,” IEEE Trans. PAMI, vol. 39, no. 6, pp. 1137–1149, June 2017.
  9. L. Deng and D. Yu, “Deep learning: Methods and applications,” Foun-dations and Trends in Signal Processing, vol. 7, no. 3–4, pp. 197–387, 2014.
  10. L. Perez and J. Wang, “The effectiveness of data augmentation in image classification using deep learning,” arXiv:1712.04621, 2017.
  11. C. Szegedy et al., “Going deeper with convolutions,” in Proc. CVPR, 2015, pp. 1–9.
  12. C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, “Rethinking the inception architecture for computer vision,” in Proc. CVPR, 2016, pp. 2818–2826.
  13. S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” in Proc. ICML, 2015, pp. 448–456.
  14. N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhut-dinov, “Dropout: A simple way to prevent neural networks from over-fitting,” J. Mach. Learn. Res., vol. 15, pp. 1929–1958, 2014.
  15. D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in Proc. ICLR, 2015.
  16. A. Vaswani et al., “Attention is all you need,” in Proc. NIPS, 2017, pp. 5998–6008.
  17. A Dosovitskiy et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” in Proc. ICLR, 2021.
  18. S. Houben et al., “Detection of traffic signs in real-world images: The German Traffic Sign Detection Benchmark,” in Proc. IJCNN, 2013, pp. 1–8.

Read more of it

Subscribe to one journal or build your own list of several. One invoice, one payment, access for a full year.

Subscribe to journals →

Publish your article

Peer-reviewed journals across every subject, with open access options. Send your manuscript to the right editor today.

Publish with us → Calculate your APC
Support