Abstract.

This paper aims to study the problem of multi-class object detection in video stream with Viola and Jones cascades. An adaptive algorithm of choosing Viola-Jones cascade based on greedy choice strategy in N-armed bandit problem is proposed. The efficiency of the algorithm is shown on the problem of detection and recognition the logo of bank card in the video stream. The proposed algorithm can be effectively used in documents localization and identification, recognition of road scene elements, search of natural and man-made objects in the coastal zone of the Arctic, and for solving other problems of rigid object detection in a heterogeneous data flows.

Keywords:

machine learning, object detection, Viola-Jones cascades, N-armed bandit problem, epsilon-greedy method, softmax method, exponential moving average.

PP. 75-82.

REFERENCES

1. Viola P., Jones M. Robust Real-time Object Detection // International Journal of Computer Vision. 2002.
2. Jones M., Viola P. Fast Multi-view Face Detection// Mitsubishi Electric Research Lab TR2000396. 2003. No. July.
3. Thompson W.R. On the likelihood that one unknown probability exceeds another in view of the evidence of two samples // Biometrika. 1933. Vol. 25, No. 3–4. P. 285–294.
4. Robbins H. Some Aspects of the Sequential Design of Experiments // Bulletin of the American Mathematical Society. 1952. Vol. 58, No. 5. P. 527–535.
5. Sutton R.S., Barto A.G. Reinforcement Learning. London: The MIT Press. 2012. 399 p.
6. Gittins J.C. Bandit processes and dynamic allocation indices // Journal of the Royal Statistical Society, Series B. 1979. Vol. 41, No 2. P. 148–177.
7. Thathachar M.A.L., Sastry P.S. A New Approach to the Design of Reinforcement Schemes for Learning Automata // IEEE Transactions on Systems, Man and Cybernetics. 1985. Vol. SMC-15, No 1. P. 168–175.
8. C. J. C. H. Watkins. Learning from Delayed Rewards. Ph.D. thesis. Cambridge University, 1989.
9. Tokic M., Palm G. Value-difference based exploration: Adaptive control between epsilon-greedy and softmax // Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2011. Vol. 7006 LNAI. P. 335–346.
10. D. Luce. Individual Choice Behavior. Wiley, 1959.
11. Grishilov A.A., Stakun V.A., Stakun A.A. Matematicheskie metody postroeniya prognozov [Mathematical methods of forecasting]. Moscow: Radio and Communications, 1997. 112 p.
12. Nikolaev D.P., Gladkov A., Chernov T., Bulatov K. Diamond recognition algorithm using two-channel x-ray radiographic separator // Proc. SPIE 9445, Seventh International Conference on Machine Vision (ICMV 2014).
13. Objed. Available at: https://github.com/usilinsergey/objed (accessed: January 1, 2017).
14. Hukovskij A. E., Tarasova N. A., Usilin S. A., Nikolaev D. P. Sintez obuchayushchej vyborki na osnove realnyh dannyh v zadachah raspoznavaniya izobrazhenij [Synthesis of the training sample based on real data in image recognition problems] // Informacionnye tekhnologii i sistemy ITiS 12 sbornik trudov konferencii [Proceedings of Information Technology and Systems (ITaS’12)]. Moscow, 2012. P. 377-382
15. Ilin D., Krivtsov V. Creating training datasets for OCR in mobile device video stream // Proceedings 29th European Conference on Modelling and Simulation. 2015.
16. Arlazarov V. V., Reshetnyak N., Slavin O. Formirovanie mnozhestva graficheskih obrazov simvolov v zadachah obucheniya klassifikatora simvolov [Forming a plurality of graphic images of characters in the character classifier learning problems] // Trudy ISA RAN [ISA RAN Proceedings]. 2014. Vol. 64. No. 4. P. 73-79.
17. Machine epsilon – Wikipedia. Available at: https://en.wikipedia.org/wiki/Machine_epsilon (accessed: September 1, 2015).
18. Precision and recall – Wikipedia. Available at: https://en.wikipedia.org/wiki/Precision_and_recall (accessed: September 1, 2015).
19. Powers D.M.W. Evaluation: From Precision, Recall and F-Factor to ROC, Informedness, Markedness & Correlation // Journal of Machine Learning Technologies. 2011. Vol 2. No 1. P. 37–63.