Joint Probabilistic Data Association using m-Best Solutions

MOT16-01 MOT16-03 MOT16-06 MOT16-07 MOT16-08 MOT16-12 MOT16-14

Short name:

JPDA_m_16

Benchmark:

Description:

In this paper, we revisit the joint probabilistic data association (JPDA) technique and propose a novel solution based on recent developments in finding the m-best solutions to an integer linear program. The key advantage of this approach is that it makes JPDA computationally tractable in applications with high target and/or clutter density, such as spot tracking in fluorescence microscopy sequences and pedestrian tracking in surveillance footage. We also show that our JPDA algorithm embedded in a simple tracking framework is surprisingly competitive with state-of-the-art global tracking methods in these two applications, while needing considerably less processing time.

The following parameter set was used for MOT16:
Prun_Thre=0.36562
tret=8.0504
Term_Frame=71.5978
PD=0.97546
q1=5.2186
Mcov=93.906
MF=1
m=100
Gatesq=20
FPPI=3
Upos=370.2145
Uvel=110.0264
AR=0.33
fpn=10

Hardware:

3 GHz, 1 CPU

Detector:

Public

Processing:

Batch

Last submitted:

March 21, 2016 (9 months ago)

Published:

March 20, 2016 at 00:00:00 CET

Submissions:

1

Open source:

Yes

Project page / code:

Reference:

H. Rezatofighi, A. Milan, Z. Zhang, Q. Shi, A. Dick, I. Reid. Joint Probabilistic Data Association Revisited. In ICCV, 2015.

Benchmark performance:

MOTAMOTPFAFMTMLFPFNID Sw.FragSpecificationsDetector
26.276.30.64.1 % 67.5 % 3,689130,5493656383 GHz, 1 CPUPublic

Detailed performance:

Sequence MOTA MOTP FAF GT MT ML FP FN ID Sw Frag
MOT16-0120.673.70.1238.7 % 56.5 % 305,042715
MOT16-0331.876.61.11485.4 % 44.6 % 1,59069,562160241
MOT16-0624.374.90.42214.5 % 70.6 % 5328,1455692
MOT16-0721.174.51.1543.7 % 70.4 % 54812,29343116
MOT16-0817.880.30.5633.2 % 68.3 % 34213,3714045
MOT16-1222.876.80.3863.5 % 69.8 % 3126,0702438
MOT16-1411.172.90.41642.4 % 82.9 % 33516,0663591

Raw data:


JPDA_m_16