TBX: Tracking-by-X


Benchmark:

Short name:

TBX

Detector:

Public

Description:

We present a novel formulation of the multiple object tracking problem which integrates low and mid-level features. In particular, we formulate the tracking problem as a quadratic program coupling detections and dense point trajectories. Due to the computational complexity of the initial QP, we propose an approximation by two auxiliary problems, a temporal and spatial association, where the temporal subproblem can be efficiently solved by a linear program and the spatial association by a clustering algorithm. The objective function of the QP is used in order to find the optimal number of clusters, where each cluster ideally represents one person. Evaluation is provided for multiple scenarios, showing the superiority of our method with respect to classic tracking-by-detection methods and also other methods that greedily integrate low-level features.

Project page / code:

Reference:

R. Henschel, L. Leal-Taixé, B. Rosenhahn, K. Schindler. Tracking with multi-level features. In arXiv:1607.07304, 2016.

Processing:

Batch

Last submitted:

March 14, 2016 (4 years ago)

Published:

November 06, 2015 at 23:40:14 CET

Submissions:

2

Open source:

No

Hardware:

3.50GHz, 1 Core

Benchmark performance:

Sequence MOTA IDF1 MOTP MT ML FP FN ID Sw.
2D MOT 201527.533.870.675.0330.07,96835,810759

Detailed performance:

Sequence MOTA IDF1 MOTP MT ML FP FN ID Sw.
ADL-Rundle-116.937.171.75.011.02,5935,08952
ADL-Rundle-329.128.872.54.011.01,2895,83785
AVG-TownCentre29.130.568.637.081.09153,956199
ETH-Crossing20.125.673.71.019.0357606
ETH-Jelmoli46.758.872.66.017.01151,22016
ETH-Linthescher16.124.274.44.0160.01317,34512
KITTI-1640.243.371.60.02.019079532
KITTI-1923.443.865.67.016.01,0462,99851
PETS09-S2L241.833.268.94.02.01,3913,949269
TUD-Crossing58.356.972.87.01.015028228
Venice-118.921.672.40.010.01133,5799

Raw data:


TBX