Method

Near Online Multi-target Tracking [NOMT]


Submitted on 21 Mar. 2015 01:57 by
Wongun Choi (NEC Laboratories)

Running time:0.09 s
Environment:16 core @ 2.5 Ghz (C++)

Method Description:
The algorithm generate consistent trajectories
given a set of detections in each frame in a (near)
online fashion. The trajectories are discovered
with a temporal latency up to 20 frames. More
details on the method will be updated
as soon as possible.
Parameters:
Reference detections are used:
http://www.cvlibs.net/download.php?
file=data_tracking_det_2.zip
Latex Bibtex:
@article{Choi2015ICCV,
author = {Choi, Wongun},
title = {Near-Online Multi-target Tracking with
Aggregated Local Flow Descriptor
},
journal= {ICCV},
year={2015},
}

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 56.49 % 52.29 % 61.59 % 54.73 % 78.75 % 64.63 % 83.40 % 81.41 %
PEDESTRIAN 36.26 % 31.87 % 41.63 % 34.61 % 61.26 % 46.88 % 62.25 % 72.83 %

Benchmark TP FP FN
CAR 23411 10981 492
PEDESTRIAN 10831 12319 2249

Benchmark MOTA MOTP MODA IDSW sMOTA
CAR 66.36 % 78.40 % 66.64 % 96 51.66 %
PEDESTRIAN 36.52 % 67.48 % 37.07 % 127 21.31 %

Benchmark MT rate PT rate ML rate FRAG
CAR 41.08 % 33.54 % 25.39 % 135
PEDESTRIAN 19.24 % 37.11 % 43.64 % 770

Benchmark # Dets # Tracks
CAR 23903 610
PEDESTRIAN 13080 256

This table as LaTeX