Multi-Camera Object Tracking (MCT) Challenge

Workshop on Visual Surveillance and Re-Identification
Zurich, 12th September 2014
In conjunction with ECCV 2014

Challenge Ranking Report

Team Ranks

 Team name  Dataset1 rank  Dataset2 rank  Dataset3 rank  Dataset4 rank  Final_score
 USC_Vision*  1  1  2  1  1.25
 hfutdspmct*  2  2  1  3  2
 CRIPAC_MCT*  3  3  3  2  2.75
 AdbTeam*  4  4  4  4  4

‘*’ denotes the results have been checked.

Team Results

USC_Vision [1,2]:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.91520  0.91322  0.51626  0.70519
 Experiment2  0.88308  0.83972  0.24265  0.43569
 Experiment3  0.59891  0.62602  0.05551  0.34043
 Subset_score  0.79906  0.79299  0.27147  0.49377
More information

hfutdspmct:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.74251  0.65441  0.73694  0.39453
 Experiment2  0.74770  0.65605  0.20277  0.26502
 Experiment3  0.28096  0.28148  0.03587  0.06077
 Subset_score  0.59039  0.53065  0.32519  0.24011
More information

CRIPAC_MCT [3]:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.66168  0.59069  0.71053  0.57031
 Experiment2  0.69032  0.62377  0.08484  0.18299
 Experiment3  0.12460  0.10746  0.01109  0.02127
 Subset_score  0.49220  0.44064  0.26882  0.25819
More information

AdbTeam:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.32036  0.34559  0.13816  0.15625
 Experiment2  0.23322  0.26679  0.11040  0.17531
 Experiment3  0.10591  0.09003  0.04056  0.04732
 Subset_score  0.21983  0.23414  0.09637  0.12629
More information

Challenge Report Overview

The main presentation of the MCT challenge can be downloaded from here.

The presentation of the winner method can be downloaded from here.

State of the Art Methods

UW_IPL [4]:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.96106  0.92647  0.78894  0.75781
 Experiment2  -  -  -  -
 Experiment3  0.60156  0.67694  0.37348  0.54287

CRF_UCRiverside [5]:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.8383  0.8015  0.6645  0.7266
 Experiment2  0.8162  0.7730  0.1240  0.4637
 Experiment3  -  -  -  -

EGTracker [6]:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.8353  0.7034  0.7417  0.3845
 Experiment2  0.8525  0.7370  0.4724  0.3778
 Experiment3  0.4120  0.4793  0.1864  0.2842

DukeMTMC [7]:

 Experiment  Dataset1 MCTA  Dataset2 MCTA  Dataset3 MCTA  Dataset4 MCTA
 Experiment1  0.7967  0.7336  0.6543  0.7616
 Experiment2  -  -  -  -
 Experiment3  -  -  -  -

Reference

[1] Chang Huang, Bo Wu and Ramakant Nevatia, "Robust object tracking by hierarchical association of detection responses," ECCV2008.

[2] Yinghao Cai and Gerard Medioni, "Exploring context information for inter-camera multiple target tracking," WACV2014.

[3] Weihua Chen, Lijun Cao, Xiaotang Chen and Kaiqi Huang, "A novel solution for multicamera object tracking," ICIP2014.

[4] Young-Gun Lee, Zheng Tang and Jenq-Neng Hwang, "Online-learning-based human tracking across non-overlapping cameras," Trans.CSVT2017.

[5] Xiaojing Chen and Bir Bhanu, "Integrating social grouping for multi-target tracking across cameras in a CRF model," Tran.CSVT2016.

[6] Weihua Chen, Lijun Cao, Xiaotang Chen and Kaiqi Huang, "An equalised global graphical model-based approach for multi-camera object tracking," Trans.CSVT2016.

[7] Ergys Ristani, Francesco Solera, Roger S. Zou,Rita Cucchiara and Carlo Tomasi, "Performance Measures and a Data Set for Multi-Target, Multi-Camera Tracking," ECCV 2016 Workshop on Benchmarking Multi-Target Tracking.