Person of Interest (Online Version)

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

Short name:

POI

Benchmark:

Description:

Faster-RCNN detector, KM and Kalman filter with Reid feature
Reid feature from:
Liu Y, Yan J, Ouyang W. Quality Aware Network for Set to Set Recognition[J]. arXiv preprint arXiv:1704.03373, 2017.

Hardware:

I7-4790, GTX970

Detector:

Private

Processing:

Online

Last submitted:

July 07, 2016 (2 years ago)

Published:

June 25, 2016 at 09:58:02 CET

Submissions:

4

Open source:

No

Project page / code:

n/a

Reference:

F. Yu, W. Li, Q. Li, Y. Liu, X. Shi, J. Yan. POI: Multiple Object Tracking with High Performance Detection and Appearance Feature. In BMTT, SenseTime Group Limited, 2016.

Benchmark performance:

MOTAMOTPFAFMTMLFPFNID Sw.FragSpecificationsDetector
66.179.50.934.0 % 20.8 % 5,06155,9148053,093I7-4790, GTX970Private
IDF1ID PrecisionID Recall
65.177.756.0

Detailed performance:

Sequence MOTA IDF1 MOTP FAF GT MT ML FP FN ID Sw Frag
MOT16-0157.756.978.41.02347.8 % 8.7 % 4452,22637114
MOT16-0378.171.779.60.914859.5 % 6.8 % 1,36421,2762741,480
MOT16-0663.660.980.70.522141.2 % 17.6 % 6403,431131284
MOT16-0757.057.079.51.05427.8 % 9.3 % 5076,42486363
MOT16-0836.243.081.40.96314.3 % 34.9 % 5849,987107302
MOT16-1248.059.079.90.68622.1 % 37.2 % 5123,75448172
MOT16-1446.257.176.71.316415.2 % 29.3 % 1,0098,816122378

Raw data:


JMC
POI