QuadMOT: Multi-Object Tracking with Quadruplet Convolutional Neural Networks

TUD-Crossing


Benchmark:

Short name:

QuadMOT

Detector:

Public

Description:

n/a

Reference:

J. Son, M. Baek, M. Cho, B. Han. Multi-Object Tracking with Quadruplet Convolutional Neural Networks. In CVPR, 2017.

Last submitted:

November 14, 2016 (3 years ago)

Published:

November 23, 2016 at 04:26:12 CET

Submissions:

2

Project page / code:

n/a

Open source:

No

Hardware:

3 GHZ, 1 Core, TITAN X

Runtime:

3.7 Hz

Benchmark performance:

Sequence MOTA IDF1 MOTP MT ML FP FN Recall Precision FAF ID Sw. Frag
2D MOT 201533.840.473.493 (12.9)266 (36.9)7,89832,06147.878.81.4703 (14.7)1,430 (29.9)

Detailed performance:

Sequence MOTA IDF1 MOTP MT ML FP FN Recall Precision FAF ID Sw. Frag
ADL-Rundle-113.635.674.0583,2074,77348.758.66.456103
ADL-Rundle-338.439.278.33113025,91141.993.40.55260
AVG-TownCentre30.849.469.841711,1913,64349.074.62.6111409
ETH-Crossing33.453.077.72154162737.590.20.205
ETH-Jelmoli42.355.475.68133791,06558.079.50.91950
ETH-Linthescher28.740.375.8141242706,05632.291.40.23881
KITTI-1638.349.971.60321980152.980.41.02960
KITTI-1933.847.968.25131,0592,41754.873.41.060191
PETS09-S2L249.028.872.6736863,94759.189.21.6285380
TUD-Crossing72.159.874.560528873.999.40.01521
Venice-131.835.772.9255392,53344.579.01.23870

Raw data: