TY - GEN
T1 - Robust online multi-object tracking by maximum a posteriori estimation with sequential trajectory prior
AU - Yang, Min
AU - Pei, Mingtao
AU - Shen, Jiajun
AU - Jia, Yunde
N1 - Publisher Copyright:
© Springer International Publishing Switzerland 2015.
PY - 2015
Y1 - 2015
N2 - This paper address the problem of online multi-object tracking by using the Maximum a Posteriori (MAP) framework. Given the observations up to the current frame, we estimate the optimal object trajectories by solving two MAP estimation problems: object detection and trajectory-detection association. By introducing the sequential trajectory prior, i.e., the prior information from previous frames about “good” trajectories, into MAP estimation, the output of the pre-trained object detector is refined and the correctness of the association between trajectories and detections is enhanced. In addition, the sequential trajectory prior allows the two MAP stages interact with each other in a sequential manner, which facilitates online multi-object tracking. Our experiments on publicly available challenging datasets demonstrate that the proposed algorithm provides superior performance in various complex scenes.
AB - This paper address the problem of online multi-object tracking by using the Maximum a Posteriori (MAP) framework. Given the observations up to the current frame, we estimate the optimal object trajectories by solving two MAP estimation problems: object detection and trajectory-detection association. By introducing the sequential trajectory prior, i.e., the prior information from previous frames about “good” trajectories, into MAP estimation, the output of the pre-trained object detector is refined and the correctness of the association between trajectories and detections is enhanced. In addition, the sequential trajectory prior allows the two MAP stages interact with each other in a sequential manner, which facilitates online multi-object tracking. Our experiments on publicly available challenging datasets demonstrate that the proposed algorithm provides superior performance in various complex scenes.
KW - Data association
KW - Maximum a posteriori estimation
KW - Online multi-object tracking
KW - Sequential trajectory prior
UR - http://www.scopus.com/inward/record.url?scp=84952795444&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-26532-2_69
DO - 10.1007/978-3-319-26532-2_69
M3 - Conference contribution
AN - SCOPUS:84952795444
SN - 9783319265315
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 623
EP - 633
BT - Neural Information Processing - 22nd International Conference, ICONIP 2015, Proceedings
A2 - Lai, Weng Kin
A2 - Liu, Qingshan
A2 - Huang, Tingwen
A2 - Arik, Sabri
PB - Springer Verlag
T2 - 22nd International Conference on Neural Information Processing, ICONIP 2015
Y2 - 9 November 2015 through 12 November 2015
ER -