Method

JRMOT [la] [on] [JRMOT]
https://github.com/StanfordVL/JRMOT_ROS

Submitted on 1 Mar. 2020 21:45 by
Mihir Patel (Stanford)

Running time:0.07 s
Environment:4 cores @ 2.5 Ghz (Python)

Method Description:
JRMOT is a novel 3D MOT system that integrates
information from 2D RGB images and 3D point
clouds into a real-time performing framework.
Our system leverages advancements in neural-
network based reidentification as well as 2D and
3D detection and descriptors. We incorporate
this into a joint probabilistic data-association
framework within a multi-modal recursive Kalman
architecture to achieve online, real-time 3D
MOT.
Parameters:
See paper.
Latex Bibtex:
@inproceedings{Shenoi2020JRMOTAR,
title = {JRMOT: A Real-Time 3D Multi-Object
Tracker and a New Large-Scale Dataset},
author = {Abhijeet Shenoi and Mihir Patel and
JunYoung Gwak and Patrick Goebel and Amir
Sadeghian and Hamid Rezatofighi and Roberto
Mart{\'i}n-Mart{\'i}n and Silvio Savarese},
year = {2020},
booktitle = {The {IEEE/RSJ} International
Conference on Intelligent Robots and Systems
({IROS})},
url={https://arxiv.org/abs/2002.08397}
}

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 69.61 % 73.05 % 66.89 % 76.95 % 85.07 % 69.18 % 88.95 % 86.72 %
PEDESTRIAN 34.24 % 38.79 % 30.55 % 42.51 % 66.64 % 32.69 % 70.12 % 76.64 %

Benchmark TP FP FN
CAR 30325 4067 787
PEDESTRIAN 12943 10207 1822

Benchmark MOTA MOTP MODA IDSW sMOTA
CAR 85.10 % 85.28 % 85.89 % 271 72.11 %
PEDESTRIAN 45.31 % 72.22 % 48.04 % 631 29.78 %

Benchmark MT rate PT rate ML rate FRAG
CAR 70.92 % 24.46 % 4.62 % 273
PEDESTRIAN 24.40 % 46.05 % 29.55 % 849

Benchmark # Dets # Tracks
CAR 31112 960
PEDESTRIAN 14765 750

This table as LaTeX