Abstract
In this paper, we analyze the existing methods of Multiple Object Tracking (MOT), point out their advantages and disadvantages. It is noted that the MOT task must be solved together with the detection of these objects, thus developing a method of the analysis of the dynamic visual scene. We propose a method of dynamic visual scene analysis based on the appearance object model. This method allows one to detect images and to get the “deep features” of detection in one Convolutional Neural Network forward pass, as well as to improve the accuracy of tracking objects construction compared to other online methods and perform processing in real time, at the speed of 24 FPS, which is shown experimentally. In addition, the method works both in the conditions of uncertainty and in the conditions of noise detection data.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsReferences
Bewley, A., Ge, Z., Ott, L., Ramos, F., Upcroft, B.: Simple online and realtime tracking. arXiv preprint arXiv: 1602.00763 (2016)
Garanin, O.I.: Method for selecting receptive field of convolutional neural network. Neurocomputers (3) 63–69 (2017)
Garanin, O.I.: Tuning method of multiscale model for detecting visual objects in a convolutional neural network. Neurocomputers (2) 50–56 (2018)
Kuhn, H.W.: The Hungarian method for the assignment problem. Nav. Res. Logist. Q. 2, 83–97 (1955)
Kim, C., Li., F.: Multiple hypothesis tracking revisited. In: Proceedings of the IEEE International Conference on Computer Vision, pp. 4696–4704 (2015)
Leal-Taixe, L., Milan, A., Reid, I., Roth, S., Schindler, K.: MOTChallenge 2015: towards a benchmark for multi-target tracking. arXiv preprint arXiv: 1504.01942 (2015)
Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S. E.: SSD: single shot multibox detector. arXiv preprint arXiv: 1512.02325 (2015)
McLaughlin, N., Rincon, J. M. D., Miller, P.: Enhancing linear programming with motion modeling for multi-target tracking. In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision, pp. 271–350 (2015)
Sadeghian, A., Alahi, A., Savarese, S.: Tracking the untrackable: learning to track multiple cues with long-term dependencies. arXiv preprint arXiv: 1701.01909 (2017)
Simonyan, K., Zisserman A.: Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv: 1409.1556 (2015)
Wojke, N., Bewley, A., Paulus, D.: Simple online and realtime tracking with a deep association metric. arXiv preprint arXiv: 1703.07402 (2017)
Acknowledgments
The work was supported by grant RFBR № 18-07-00928_a “Methods and technologies of intelligent support for research of complex hydro-mechanical processes in conditions of uncertainty on the convoluted neuro-fuzzy networks”.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer Nature Switzerland AG
About this paper
Cite this paper
Borisov, V.V., Garanin, O.I. (2018). A Method of Dynamic Visual Scene Analysis Based on Convolutional Neural Network. In: Kuznetsov, S., Osipov, G., Stefanuk, V. (eds) Artificial Intelligence. RCAI 2018. Communications in Computer and Information Science, vol 934. Springer, Cham. https://doi.org/10.1007/978-3-030-00617-4_6
Download citation
DOI: https://doi.org/10.1007/978-3-030-00617-4_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-00616-7
Online ISBN: 978-3-030-00617-4
eBook Packages: Computer ScienceComputer Science (R0)