Abstract. 

In this work we study a ID document recognition task. We propose an approach based on VGG network variants, which allows to get the most expressive docmument descriptors for this task. We also propose a combined neural network architecture effective for document verification. The proposed architecture includes auto-encoder generating an expressive image descriptor. This apphiach we compare to the existing analogues, and provide results of evaluation on specific dataset of ID documents owned by VK.com.

Keywords: 

optical character recognition, image classification, noise robustness, neural networks.

PP. 15-23.

DOI: 10.14357/20790279200102

 

References

1. Javidi B., Horner J.L. Optical pattern recognition for validation and security verification //Optical engineering. – 1994. – Т. 33. – №. 6. – С. 1752-1757.

2. Visilter Y.V., Zheltov S.Y., Lukin A.A. Development of OCR system for portable passport and visa reader //Document Recognition and Retrieval VI.– International Society for Optics and Photonics, 1999. – Т. 3651. – С. 194-199.

3. Limonova E. et al. Slant rectification in Russian passport OCR system using fast Hough transform // Ninth International Conference on Machine Vision (ICMV 2016). – International Society for Optics and Photonics, 2017. – Т. 10341. – С. 103410P.

4. Kim K.B., Oh A., Woo Y.W. PCA-based face verification and passport code recognition using improved FKCN algorithm //2008 Eighth International Conference on Intelligent Systems Design and Applications. – IEEE, 2008. – Т. 2. – С. 51-57.

5. Kim K.B., Kim S. A passport recognition and face verification using enhanced fuzzy ART based RBF network and PCA algorithm //Neurocomputing. – 2008. – Т. 71. – №. 16-18. – С. 3202-3210.

6. Simonyan K., Zisserman A. Very deep convolutional networks for large-scale image recognition //arXiv preprint arXiv:1409.1556. – 2014.

7. Szegedy C. et al. Going deeper with convolutions // Proceedings of the IEEE conference on computervision and pattern recognition. – 2015. – С. 1-9.

8. He K. et al. Deep residual learning for image recognition //Proceedings of the IEEE conference on computer vision and pattern recognition. – 2016. – С. 770-778.

9. Howard A.G. et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications //arXiv preprint arXiv:1704.04861. – 2017.

10. Tianyi Liu, Shuangsang Fang, Yuehui Zhao, Peng Wang, and Jun Zhang. Implementation of training convolutional neural networks //CoRR, abs/1506.01195. - 2015.

11. Deng J. et al. Imagenet: A large-scale hierarchical image database //2009 IEEE conference on computer vision and pattern recognition. – Ieee, 2009. – С. 248-255.

12. Hinton G.E., Salakhutdinov R.R. Reducing the dimensionality of data with neural networks // Science (New York, NY). – 2006. – Т. 313. – № 5786. – С. 504-507.

13. Gupta S., Mazumdar S.G. Sobel edge detection algorithm //International journal of computer science and management Research. – 2013. – Т. 2. – №. 2. – С. 1578-1583.

14. Elboher E., Werman M. Efficient and accurate Gaussian image filtering using running sums //2012 12th International Conference on Intelligent Systems Design and Applications (ISDA). – IEEE, 2012. – С. 897-902.

15. Metz J. et al. Comparing published multi-label classifier performance measures to the ones obtained by a simple multi-label baseline classifier //arXiv preprint arXiv:1503.06952. – 2015.