AI Paper Atlas

Research map / Detection and segmentation / Tracking

Track Without Appearance: Learn Box and Tracklet Embedding With Local and Global Motion Patterns for Vehicle Tracking

Gaoang Wang, Renshu Gu, Zuozhu Liu, Weijie Hu, Mingli Song, Jenq-Neng Hwang

ICCV 2021 · Detection and segmentation › Tracking · approach: Memory and correlation-based MOT · 78 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 binary cross-entropy loss to triplet loss for box embedding (ablation)LGM box embedding · KITTI-car tracking validationHOTA: 75.7 -> 76.9Improved
add binary cross-entropy loss to triplet loss for box embedding (ablation)LGM box embedding · KITTI-car tracking validationMOTA: 88.0 -> 89.0Improved
add binary cross-entropy loss to tracklet embedding loss (ablation)LGM tracklet embedding · KITTI-car tracking validationHOTA: 76.3 -> 76.9Improved
add binary cross-entropy loss to tracklet embedding loss (ablation)LGM tracklet embedding · KITTI-car tracking validationMOTA: 88.2 -> 89.0Improved
add reconstruction loss to tracklet embedding training (ablation)LGM tracklet embedding · KITTI-car tracking validationconvergenceImproved
add tracklet embedding module to box embedding (ablation)LGM · KITTI-car tracking validationHOTA: 75.2 -> 76.9Improved
add tracklet embedding module to box embedding (ablation)LGM · KITTI-car tracking validationAssA: 74.3 -> 77.2Improved
add tracklet embedding module to box embedding (ablation)LGM · KITTI-car tracking validationMOTA: 88.0 -> 89.0Improved
use LGM motion-only tracker instead of appearance-based CenterTrackvehicle tracking · KITTI-car tracking testMOTA: 88.8 -> 87.6Hurt
use LGM motion-only tracker instead of appearance-based CenterTrackvehicle tracking · KITTI-car tracking testHOTA: 73.0 -> 73.1No clear effect
use LGM instead of best prior method FAMNetvehicle tracking · UA-Detrac testPR-MOTA: 19.8 -> 22.5Improved
use LGM instead of IOU Tracker on pedestriansmotion-only pedestrian tracking · MOT17 pedestrian testMOTA: 45.5 -> 56.0Improved

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