Method

Multi-target Tracking Algorithm Based on Fusion and NC2 Noise Covariance Estimation [la][on] [Opm-NC2]


Submitted on 28 Jun. 2021 09:44 by
Jiang Chao (University of science and technology of China)

Running time:0.01 s
Environment:1 core @ 3.0 Ghz (C/C++)

Method Description:
Laser radar data is prone to false detection
because of its simple features. To solve this
problem, we propose a fast image-based re-
classification algorithm. Each detection result of
laser radar is classified and recognized by the re-
classification algorithm, so as to reduce the false
detection. Then,the detection results are fed into
our proposed NC2 adaptive filter and multi-target
tracker for multi-target tracking.
Parameters:
python3.7
Latex Bibtex:
@InProceedings{2022_IEEE sensors journal,
author={Chao Jiang, Zhiling Wang, Huawei Liang},
title={A Fast and High-Performance Object Proposal
Method for Vision Sensors: Application to Object
Detection},
booktitle = {IEEE sensors journal},
month = {January},
year = {2022}
}

Detailed Results

From all 29 test sequences, our benchmark computes the HOTA tracking metrics (HOTA, DetA, AssA, DetRe, DetPr, AssRe, AssPr, LocA) [1] as well as the CLEARMOT, MT/PT/ML, identity switches, and fragmentation [2,3] metrics. The tables below show all of these metrics.


Benchmark HOTA DetA AssA DetRe DetPr AssRe AssPr LocA
CAR 73.19 % 73.27 % 73.77 % 80.98 % 81.67 % 77.05 % 89.84 % 87.31 %
PEDESTRIAN 46.55 % 46.82 % 46.68 % 53.01 % 59.38 % 50.84 % 65.82 % 72.07 %

Benchmark TP FP FN
CAR 31629 2763 2472
PEDESTRIAN 16990 6160 3679

Benchmark MOTA MOTP MODA IDSW sMOTA
CAR 84.21 % 85.86 % 84.78 % 195 71.20 %
PEDESTRIAN 56.05 % 65.68 % 57.50 % 335 30.86 %

Benchmark MT rate PT rate ML rate FRAG
CAR 75.85 % 18.15 % 6.00 % 301
PEDESTRIAN 44.33 % 43.30 % 12.37 % 1432

Benchmark # Dets # Tracks
CAR 34101 1085
PEDESTRIAN 20669 768

This table as LaTeX