Ensemble-Based Tracking: Aggregating Crowdsourced Structured Time Series Data

Naiyan Wang, Dit-Yan Yeung
Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):1107-1115, 2014.

Abstract

We study the problem of aggregating the contributions of multiple contributors in a crowdsourcing setting. The data involved is in a form not typically considered in most crowdsourcing tasks, in that the data is structured and has a temporal dimension. In particular, we study the visual tracking problem in which the unknown data to be estimated is in the form of a sequence of bounding boxes representing the trajectory of the target object being tracked. We propose a factorial hidden Markov model (FHMM) for ensemble-based tracking by learning jointly the unknown trajectory of the target and the reliability of each tracker in the ensemble. For efficient online inference of the FHMM, we devise a conditional particle filter algorithm by exploiting the structure of the joint posterior distribution of the hidden variables. Using the largest open benchmark for visual tracking, we empirically compare two ensemble methods constructed from five state-of-the-art trackers with the individual trackers. The promising experimental results provide empirical evidence for our ensemble approach to "get the best of all worlds".

Cite this Paper

BibTeX
@InProceedings{pmlr-v32-wangg14, title = {Ensemble-Based Tracking: Aggregating Crowdsourced Structured Time Series Data}, author = {Wang, Naiyan and Yeung, Dit-Yan}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {1107--1115}, year = {2014}, editor = {Xing, Eric P. and Jebara, Tony}, volume = {32}, number = {2}, series = {Proceedings of Machine Learning Research}, address = {Bejing, China}, month = {22--24 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v32/wangg14.pdf}, url = {https://proceedings.mlr.press/v32/wangg14.html}, abstract = {We study the problem of aggregating the contributions of multiple contributors in a crowdsourcing setting. The data involved is in a form not typically considered in most crowdsourcing tasks, in that the data is structured and has a temporal dimension. In particular, we study the visual tracking problem in which the unknown data to be estimated is in the form of a sequence of bounding boxes representing the trajectory of the target object being tracked. We propose a factorial hidden Markov model (FHMM) for ensemble-based tracking by learning jointly the unknown trajectory of the target and the reliability of each tracker in the ensemble. For efficient online inference of the FHMM, we devise a conditional particle filter algorithm by exploiting the structure of the joint posterior distribution of the hidden variables. Using the largest open benchmark for visual tracking, we empirically compare two ensemble methods constructed from five state-of-the-art trackers with the individual trackers. The promising experimental results provide empirical evidence for our ensemble approach to "get the best of all worlds".} }
Endnote
%0 Conference Paper %T Ensemble-Based Tracking: Aggregating Crowdsourced Structured Time Series Data %A Naiyan Wang %A Dit-Yan Yeung %B Proceedings of the 31st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2014 %E Eric P. Xing %E Tony Jebara %F pmlr-v32-wangg14 %I PMLR %P 1107--1115 %U https://proceedings.mlr.press/v32/wangg14.html %V 32 %N 2 %X We study the problem of aggregating the contributions of multiple contributors in a crowdsourcing setting. The data involved is in a form not typically considered in most crowdsourcing tasks, in that the data is structured and has a temporal dimension. In particular, we study the visual tracking problem in which the unknown data to be estimated is in the form of a sequence of bounding boxes representing the trajectory of the target object being tracked. We propose a factorial hidden Markov model (FHMM) for ensemble-based tracking by learning jointly the unknown trajectory of the target and the reliability of each tracker in the ensemble. For efficient online inference of the FHMM, we devise a conditional particle filter algorithm by exploiting the structure of the joint posterior distribution of the hidden variables. Using the largest open benchmark for visual tracking, we empirically compare two ensemble methods constructed from five state-of-the-art trackers with the individual trackers. The promising experimental results provide empirical evidence for our ensemble approach to "get the best of all worlds".
RIS
TY - CPAPER TI - Ensemble-Based Tracking: Aggregating Crowdsourced Structured Time Series Data AU - Naiyan Wang AU - Dit-Yan Yeung BT - Proceedings of the 31st International Conference on Machine Learning DA - 2014/06/18 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-wangg14 PB - PMLR DP - Proceedings of Machine Learning Research VL - 32 IS - 2 SP - 1107 EP - 1115 L1 - http://proceedings.mlr.press/v32/wangg14.pdf UR - https://proceedings.mlr.press/v32/wangg14.html AB - We study the problem of aggregating the contributions of multiple contributors in a crowdsourcing setting. The data involved is in a form not typically considered in most crowdsourcing tasks, in that the data is structured and has a temporal dimension. In particular, we study the visual tracking problem in which the unknown data to be estimated is in the form of a sequence of bounding boxes representing the trajectory of the target object being tracked. We propose a factorial hidden Markov model (FHMM) for ensemble-based tracking by learning jointly the unknown trajectory of the target and the reliability of each tracker in the ensemble. For efficient online inference of the FHMM, we devise a conditional particle filter algorithm by exploiting the structure of the joint posterior distribution of the hidden variables. Using the largest open benchmark for visual tracking, we empirically compare two ensemble methods constructed from five state-of-the-art trackers with the individual trackers. The promising experimental results provide empirical evidence for our ensemble approach to "get the best of all worlds". ER -
APA
Wang, N. & Yeung, D.. (2014). Ensemble-Based Tracking: Aggregating Crowdsourced Structured Time Series Data. Proceedings of the 31st International Conference on Machine Learning, in Proceedings of Machine Learning Research 32(2):1107-1115 Available from https://proceedings.mlr.press/v32/wangg14.html.

Related Material