ApLift: Approximate Lifted Disjoint Paths Tracker

MOT20-04


Benchmark:

Short name:

ApLift

Detector:

Public

Description:

Reference:

A. Hornakova*, T. Kaiser*, M. Rolinek, B. Rosenhahn, P. Swoboda, R. Henschel. Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths. In International Conference on Computer Vision (ICCV), 2021.

Last submitted:

March 17, 2021 (6 months ago)

Published:

March 22, 2021 at 11:46:14 CET

Submissions:

3

Project page / code:

Open source:

No

Hardware:

5 GHZ 8 Core

Runtime:

0.4 Hz

Benchmark performance:

Sequence MOTA IDF1 HOTA MT ML FP FN Rcll Prcn AssA DetA AssRe AssPr DetRe DetPr LocA FAF ID Sw. Frag
MOT2058.956.546.6513 (41.3)264 (21.3)17,739192,73662.894.845.248.248.176.851.377.582.24.02,241 (0.0)2,112 (0.0)

Detailed performance:

Sequence MOTA IDF1 HOTA MT ML FP FN Rcll Prcn AssA DetA AssRe AssPr DetRe DetPr LocA FAF ID Sw. Frag
MOT20-0479.368.856.6412408,31547,36482.796.550.563.753.778.867.979.282.74.0968840
MOT20-0636.136.830.1411114,78679,31340.391.829.630.831.571.432.574.180.64.7740744
MOT20-0756.954.744.0401593613,13560.395.542.246.445.375.949.278.082.21.6194195
MOT20-0826.533.827.620983,70252,92431.786.931.824.134.370.025.570.080.14.6339333

Raw data: