LP_SSVM: Linear Programming Learned With Structured SVM

TUD-Crossing


Benchmark:

Short name:

LP_SSVM

Detector:

Public

Description:

Min-cost network flow model with pairwise interactions; network cost is learned using structured SVM.

Project page / code:

n/a

Reference:

S. Wang, C. Fowlkes. Learning Optimal Parameters for Multi-target Tracking with Contextual Interactions. In International Journal of Computer Vision, 2016.

Processing:

Batch

Last submitted:

April 20, 2015 (5 years ago)

Published:

April 21, 2015 at 04:22:07 CET

Submissions:

1

Open source:

No

Hardware:

2.4GHz, 1 Core

Runtime:

41.3 Hz

Benchmark performance:

Sequence MOTA IDF1 MOTP MT ML FP FN Recall Precision FAF ID Sw. Frag
2D MOT 201525.234.071.742 (5.8)382 (53.0)8,36936,93239.974.51.4646 (16.2)849 (21.3)

Detailed performance:

Sequence MOTA IDF1 MOTP MT ML FP FN Recall Precision FAF ID Sw. Frag
ADL-Rundle-114.038.371.9773,5074,42352.558.27.06992
ADL-Rundle-328.036.972.94101,8555,38847.072.03.08183
AVG-TownCentre14.724.370.161394595,51522.878.11.0123141
ETH-Crossing24.937.575.61171074126.196.30.022
ETH-Jelmoli39.552.474.48162241,29349.084.70.51729
ETH-Linthescher15.623.575.65157417,48316.297.20.01115
KITTI-1639.241.773.6029092445.789.60.42029
KITTI-1928.239.665.94188102,95544.774.70.870162
PETS09-S2L241.527.970.5376294,80350.288.51.4212249
TUD-Crossing60.055.074.2424837566.093.80.21820
Venice-117.831.973.0076963,03233.668.71.52327

Raw data: