Multi-target tracking algorithm based on deep learning

Abstract With the continuous development of deep learning in multi-target tracking, the use of convolutional neural network for feature extraction has replaced the traditional feature extraction method, but the accuracy of target tracking needs to be improved. In order to further improve the accurac...

Full description

Saved in:
Bibliographic Details
Published inJournal of physics. Conference series Vol. 1948; no. 1; pp. 12011 - 12015
Main Authors Huo, Wenxiao, Ou, Jiayu, Li, Tianping
Format Journal Article
LanguageEnglish
Published Bristol IOP Publishing 01.06.2021
Subjects
Accuracy
Algorithms
Artificial neural networks
Deep learning
Feature extraction
Machine learning
Multiple target tracking
Physics
Sonar
Target detection
Tracking
Online AccessGet full text

Cover

More Information
Summary:Abstract With the continuous development of deep learning in multi-target tracking, the use of convolutional neural network for feature extraction has replaced the traditional feature extraction method, but the accuracy of target tracking needs to be improved. In order to further improve the accuracy of multi-target tracking, a new multi-target tracking algorithm based on RFB is proposed in this paper. The algorithm is mainly divided into three parts: multi-target detection, feature extraction and multi-target tracking. In the multi-target detection part, CenterNet was selected as the detection network to improve detection accuracy. In feature extraction part, RFBNET is combined with pedestrian re-recognition network to strengthen feature extraction capability. DeepSORT algorithm is used in multi - target tracking. Experimental results on MOT16 data set show that the proposed algorithm is more effective than other methods.
ISSN:1742-6588
1742-6596
DOI:10.1088/1742-6596/1948/1/012011