Learning an image-based motion context for multi-target tracking

TUD-Crossing PETS09-S2L2 ETH-Jelmoli ETH-Linthescher ETH-Crossing AVG-TownCentre ADL-Rundle-1 ADL-Rundle-3 KITTI-16 KITTI-19

Short name:

MotiCon

Benchmark:

Description:

We present a novel method for multiple people tracking that leverages a generalized model for capturing interactions among individuals.
At the core of our model lies a learned dictionary of interaction feature strings which capture relationships between the motions of targets.
These feature strings, created from low-level image features, lead to a much richer representation of the physical interactions between targets compared to hand-specified social force models that previous works have introduced for tracking. One disadvantage of using social forces is that all pedestrians must be detected in order for the forces to be applied, while our method is able to encode the effect of undetected targets, making the tracker more robust to partial occlusions.
The interaction feature strings are used in a Random Forest framework to track targets according to the features surrounding them.

Hardware:

2.6 GHz, 1 core

Detector:

Public

Processing:

Batch

Last submitted:

April 07, 2015 (2 years ago)

Published:

April 07, 2015 at 21:02:03 CET

Submissions:

1

Open source:

Yes

Project page / code:

Reference:

L. Leal-Taixé, M. Fenzi, A. Kuznetsova, B. Rosenhahn, S. Savarese. Learning an image-based motion context for multiple people tracking. In CVPR, 2014.

Benchmark performance:

MOTAMOTPFAFMTMLFPFNID Sw.FragSpecificationsDetector
23.170.91.84.7 % 52.0 % 10,40435,8441,0181,0612.6 GHz, 1 corePublic
IDF1ID PrecisionID Recall
29.439.823.3

Detailed performance:

Sequence MOTA IDF1 MOTP FAF GT MT ML FP FN ID Sw Frag
TUD-Crossing58.246.670.80.21323.1 % 15.4 % 324032632
PETS09-S2L246.627.267.61.3429.5 % 14.3 % 5604,354238264
ETH-Jelmoli43.552.472.90.74520.0 % 28.9 % 2951,1023751
ETH-Linthescher18.324.077.70.11971.5 % 74.1 % 987,1247275
ETH-Crossing22.829.172.90.1263.8 % 65.4 % 1375388
AVG-TownCentre11.923.070.30.82260.9 % 69.9 % 3535,8727475
ADL-Rundle-11.030.470.38.93218.8 % 12.5 % 4,4494,628136170
ADL-Rundle-318.122.971.84.4444.5 % 20.5 % 2,7555,355217140
KITTI-1638.839.570.10.7170.0 % 11.8 % 1428633648
KITTI-1933.841.769.90.8626.5 % 21.0 % 8872,552100126
Venice-118.226.172.91.8170.0 % 29.4 % 8202,8387472

Raw data:


MDP
MotiCon