Method

QD-3DT [on] [QD-3DT]
https://eborboihuc.github.io/QD-3DT

Submitted on 7 Dec. 2020 20:25 by
Hou-Ning Hu (Berkeley )

Running time:0.03 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
End-to-end 3D detection and tracking
Parameters:
None
Latex Bibtex:
@article{Hu2021QD3DT,
author = {Hu, Hou-Ning and Yang, Yung-Hsu and
Fischer, Tobias and Yu, Fisher and Darrell, Trevor
and Sun, Min},
title = {Monocular Quasi-Dense 3D Object Tracking},
journal = {ArXiv:2103.07351},
year = {2021}
}

Detailed Results

From all 29 test sequences, our benchmark computes the commonly used tracking metrics CLEARMOT, MT/PT/ML, identity switches, and fragmentations [1,2]. The tables below show all of these metrics.


Benchmark MOTA MOTP MODA MODP
CAR 86.41 % 85.82 % 86.73 % 88.70 %
PEDESTRIAN 52.98 % 73.41 % 55.09 % 91.61 %

Benchmark recall precision F1 TP FP FN FAR #objects #trajectories
CAR 90.44 % 97.79 % 93.97 % 35588 804 3761 7.23 % 40958 1450
PEDESTRIAN 64.49 % 87.86 % 74.38 % 15091 2086 8311 18.75 % 19571 1436

Benchmark MT PT ML IDS FRAG
CAR 75.38 % 22.15 % 2.46 % 108 553
PEDESTRIAN 32.30 % 49.14 % 18.56 % 488 1393

This table as LaTeX


[1] K. Bernardin, R. Stiefelhagen: Evaluating Multiple Object Tracking Performance: The CLEAR MOT Metrics. JIVP 2008.
[2] Y. Li, C. Huang, R. Nevatia: Learning to associate: HybridBoosted multi-target tracker for crowded scene. CVPR 2009.


eXTReMe Tracker