AI Paper Atlas

Research map / Detection and segmentation / CNN features

FairMOT: On the Fairness of Detection and Re-identification in Multiple Object Tracking

Yifu Zhang, Chunyu Wang, Xinggang Wang, Wenjun Zeng, Wenyu Liu

IJCV 2021 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 2,059 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
sample re-ID features at object centers instead of ROI-Align (ablation)FairMOT · MOT17 valIDF1: 71.0 -> 72.8Improved
use one-stage center sampling instead of two-stage re-ID (ablation)FairMOT · MOT17 valIDF1: 68.2 -> 72.8Improved
add bi-linear interpolation to center sampling (ablation)FairMOT · MOT17 valIDF1 / MOTA: IDF1 72.8 -> 74.3, MOTA 69.1 -> 68.8Mixed
replace fixed loss weights with uncertainty weighting (ablation)FairMOT · MOT17 valIDs / MOTA: IDs 387 -> 299, MOTA 69.6 -> 69.1Mixed
add multi-layer feature fusion (FPN) to RegNetY-4.0GF (ablation)FairMOT · MOT17 valMOTA: 63.9 -> 65.8Improved
use larger backbone ResNet-50 instead of ResNet-34 (ablation)FairMOT · MOT17 valMOTA: 63.6 -> 63.7No clear effect
add FPN fusion to ResNet-34 (ablation)FairMOT · MOT17 valTPR: 90.9 -> 94.2Improved
keep deformable convolutions in DLA-34 (ablation)FairMOT · MOT17 valMOTA: 65.0 -> 69.1Improved
train the re-ID branch alongside detection (feature conflict) (ablation)FairMOT ResNet-34 · MOT17 valAP: 76.1 -> 75.1Hurt
reduce re-ID feature dimension from 512 to 64 (ablation)FairMOT · MOT17 valMOTA: 68.5 -> 69.2Improved
reduce re-ID feature dimension from 512 to 64 (ablation)JDE · MOT17 valMOTA: 59.9 -> 60.3Improved
add Kalman filter to re-ID association (ablation)FairMOT DLA-34 · MOT17 valIDs: 435 -> 342Improved

Later papers that compared with this method

14 later accepted papers report FairMOT as a baseline in their results tables; 11 of them report a better number on most of the results they share with it.

6 more not listed.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy