Multiple Target Tracking by Learning Feature Representation and Distance Metric Jointly

MOT16-01 MOT16-03 MOT16-06 MOT16-07 MOT16-08 MOT16-12 MOT16-14

Short name:

TripT

Benchmark:

Description:

This algorithm proposes a novel affinity model by learning feature representation and distance metric jointly in a unified deep architecture.

Hardware:

2.10GHz

Detector:

Public

Processing:

Online

Last submitted:

March 24, 2018 (27 days ago)

Published:

March 24, 2018 at 09:04:31 CET

Submissions:

1

Open source:

No

Project page / code:

n/a

Reference:

Anonymous submission

Benchmark performance:

MOTAMOTPFAFMTMLFPFNID Sw.FragSpecificationsDetector
44.374.30.512.5 % 46.5 % 2,79798,3324691,1342.10GHzPublic
IDF1ID PrecisionID Recall
45.871.133.8

Detailed performance:

Sequence MOTA IDF1 MOTP FAF GT MT ML FP FN ID Sw Frag
MOT16-0140.857.870.90.12321.7 % 43.5 % 433,740121
MOT16-0350.643.774.41.014817.6 % 21.6 % 1,51349,817355637
MOT16-0645.154.872.30.222110.9 % 50.7 % 2316,08022135
MOT16-0737.850.572.90.65411.1 % 38.9 % 3009,81936142
MOT16-0835.047.478.40.36322.2 % 46.0 % 21110,6412142
MOT16-1244.259.376.10.28614.0 % 48.8 % 1584,461735
MOT16-1423.536.172.80.51644.9 % 65.2 % 34113,77427122

Raw data:

n/a


TripT