Multi Object Tracking via Dynamic Programming on min cost flow networks augmented by YOLO detections

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

Short name:

DP_YOLO

Benchmark:

Description:

An improved version of the DP+NMS tracker.
However, this does not use NMS and the detections are replaced by YOLO.
Skip connections in the graph will also be used.

Hardware:

Intel i7 1 Core

Detector:

Private

Processing:

Batch

Last submitted:

April 02, 2019 (7 months ago)

Published:

March 18, 2019 at 07:45:39 CET

Submissions:

3

Open source:

Yes

Project page / code:

Reference:

Anonymous submission

Benchmark performance:

MOTAMOTPFAFMTMLFPFNID Sw.FragSpecificationsDetector
37.474.81.019.6 % 43.3 % 5,56932,529394808Intel i7 1 CorePrivate
IDF1ID PrecisionID Recall
42.158.632.9

Detailed performance:

Sequence MOTA IDF1 MOTP FAF GT MT ML FP FN ID Sw Frag
TUD-Crossing76.458.476.20.21369.2 % 0.0 % 382061619
PETS09-S2L244.730.472.81.1427.1 % 19.0 % 4664,740125244
ETH-Jelmoli34.257.075.81.94531.1 % 26.7 % 8498061546
ETH-Linthescher57.958.777.30.819732.0 % 40.1 % 8972,8016596
ETH-Crossing63.054.080.90.32626.9 % 38.5 % 69296611
AVG-TownCentre25.941.970.11.322610.2 % 55.3 % 6024,66830111
ADL-Rundle-133.141.674.21.63221.9 % 34.4 % 8215,3792766
ADL-Rundle-338.342.978.11.74427.3 % 25.0 % 1,0595,1566074
KITTI-1614.324.167.60.1170.0 % 76.5 % 251,433020
KITTI-1916.926.967.50.2621.6 % 61.3 % 2254,1902372
Venice-125.530.172.41.21711.8 % 29.4 % 5182,8542749

Raw data:

n/a


DP_YOLO