AI Paper Atlas

Research map / Detection and segmentation / Tracking

Lost and Found: Overcoming Detector Failures in Online Multi-Object Tracking

Lorenzo Vaquero, Yihong Xu, Xavier Alameda-Pineda, Victor M. Brea, Manuel Mucientes

ECCV 2024 · Detection and segmentation › Tracking · approach: Association cues for online MOT · 27 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
add BUSCA to online trackersTransCenter · MOT17 testHOTA: 56.6 -> 61.7Improved
add BUSCA to CenterTrackCenterTrack · MOT20 testIDF1: 36.6 -> 58.0Improved
add BUSCA to ByteTrackByteTrack · MOT17 testMOTA: 78.9 -> 79.3Improved
add BUSCA to GHOSTGHOST · MOT17 testIDF1: 77.1 -> 77.0No clear effect
add BUSCA to TransCenter on MOT17 test MOTATransCenter · MOT17 testMOTA: 76.2 -> 76.2No clear effect
lower detection threshold to 0.01 before association (ablation)ByteTrack · MOT17 valMOTA: 76.5 -> 75.3Hurt
add extra Kalman IoU recovery association round (ablation)ByteTrack · MOT17 valMOTA: 76.5 -> 75.4Hurt
add Mixed IoU and appearance recovery round (ablation)ByteTrack · MOT17 valMOTA: 76.5 -> 76.6No clear effect
use only [Halluc.] learned token (ablation)BUSCA · MOT17 valMOTA: 76.5 -> 75.0Hurt
combine [Halluc.] and [Miss.] tokens (ablation)BUSCA · MOT17 valMOTA: 75.0 -> 76.5Improved
add spatiotemporal encoding (STE) (ablation)BUSCA · MOT17 valMOTA: 76.5 -> 76.7Improved
add contextual proposals from neighbors (ablation)BUSCA · MOT17 valMOTA: 76.7 -> 77.1Improved

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