Method

Near-Online Multi-target Tracking (HM baseline) [on] [NOMT-HM]


Submitted on 16 Oct. 2014 23:37 by
Wongun Choi (NEC Laboratories)

Running time:0.09 s
Environment:8 cores @ 2.5 Ghz (Matlab + C/C++)

Method Description:
The algorithm tracks targets using Hungarian
method equipped with the same match features used
in NOMT method. It makes the association in every
time frame in pure online fashion. More details on
the method will be updated as soon as possible.
Parameters:
Reference detections are used:
http://www.cvlibs.net/download.php?
file=data_tracking_det_2.zip
Latex Bibtex:
@article{Choi2015ICCV,
author = {Choi, Wongun},
title = {Near-Online Multi-target Tracking with
Aggregated Local Flow Descriptor
},
journal= {ICCV},
year={2015},
}

Detailed Results

From all 29 test sequences, our benchmark computes the commonly used tracking metrics CLEARMOT, MT/PT/ML, identity switches, and fragmentations [1,2]. The tables below show all of these metrics.


Benchmark MOTA MOTP MODA MODP
CAR 61.17 % 78.65 % 61.26 % 84.15 %
PEDESTRIAN 27.49 % 67.99 % 27.81 % 93.02 %

Benchmark recall precision F1 TP FP FN FAR #objects #trajectories
CAR 64.30 % 97.83 % 77.60 % 23075 512 12813 4.60 % 26309 865
PEDESTRIAN 37.30 % 80.09 % 50.89 % 8660 2153 14559 19.35 % 13197 405

Benchmark MT PT ML IDS FRAG
CAR 33.85 % 38.15 % 28.00 % 28 241
PEDESTRIAN 15.12 % 34.36 % 50.52 % 73 732

This table as LaTeX


[1] K. Bernardin, R. Stiefelhagen: Evaluating Multiple Object Tracking Performance: The CLEAR MOT Metrics. JIVP 2008.
[2] Y. Li, C. Huang, R. Nevatia: Learning to associate: HybridBoosted multi-target tracker for crowded scene. CVPR 2009.


eXTReMe Tracker