Method

CenterTube-V [CenterTube-V]
[Anonymous Submission]

Submitted on 20 Oct. 2021 09:28 by
[Anonymous Submission]

Running time:0.14 s
Environment:1 core @ 2.5 Ghz (Python)

Method Description:
3D Multi-Object Tracking (MOT) in dynamic point
cloud sequences is a crucial component for
numerous intelligent systems such as autonomous
driving and robotics. We propose a concise one-
step model CenterTube by formulating the
problem of target trajectory prediction as 4D
tubelet detection in a short video clip. Our
CenterTube is inspired by the anchor-free 3D
detector CenterPoint and consists of three head
branches: (1) a Center Branch to detect the
center points of objects and recognize their
category in key frame; (2) a Box Branch to
regress 3D size and orientation of bounding
box; (3) a Movement Branch to estimate instance
movement and frame interval. Further, a Tube
BEV-IoU (TB-IoU) is proposed to link the
generated clip-level tubelets and form the
final tracks.
Parameters:
python=3.6
Latex Bibtex:

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 67.76 % 69.95 % 66.96 % 76.09 % 81.17 % 72.46 % 80.97 % 85.67 %

Benchmark TP FP FN
CAR 30515 3877 1727

Benchmark MOTA MOTP MODA IDSW sMOTA
CAR 81.56 % 83.86 % 83.71 % 737 67.24 %

Benchmark MT rate PT rate ML rate FRAG
CAR 69.85 % 21.54 % 8.62 % 589

Benchmark # Dets # Tracks
CAR 32242 714

This table as LaTeX