MOTDT: Real-time Multiple People Tracking with Deeply Learned Candidate Selection and Person Re-identification

MOT16-01


Benchmark:

MOT16 |

Short name:

MOTDT

Detector:

Public

Description:

Real-time Multiple People Tracking with Deeply Learned Candidate Selection and Person Re-Identification

Reference:

Last submitted:

October 30, 2017 (2 years ago)

Published:

July 06, 2018 at 09:12:13 CET

Submissions:

2

Project page / code:

Open source:

Yes

Hardware:

GTX 1080, 2.3 GHz, 1 Core

Runtime:

20.6 Hz

Benchmark performance:

Sequence MOTA IDF1 MOTP MT ML FP FN Recall Precision FAF ID Sw. Frag
MOT1647.650.974.8115 (15.2)291 (38.3)9,25385,43153.191.31.6792 (14.9)1,858 (35.0)

Detailed performance:

Sequence MOTA IDF1 MOTP MT ML FP FN Recall Precision FAF ID Sw. Frag
MOT16-0140.142.871.3472293,56444.392.50.53577
MOT16-0355.454.774.839235,28840,93760.892.33.5361848
MOT16-0645.151.373.634985875,66750.990.90.582219
MOT16-0744.148.673.66166448,37448.792.51.3101263
MOT16-0834.639.579.0102366910,17339.290.81.1104143
MOT16-1238.951.776.312357684,26248.684.00.93576
MOT16-1426.441.174.010891,06812,45432.685.01.474232

Raw data: