Research map / Detection and segmentation / Tracking
Large Scale Real-World Multi-person Tracking
Bing Shuai, Alessandro Bergamo, Uta Büchler, Andrew Berneshawi, Alyssa Boden, Joseph Tighe
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 changed | Where | Measured | Effect |
|---|---|---|---|
| apply occlusion filter (exclude fully occluded boxes) at evaluation | SiamMOT · PersonPath22 test | MOTA: 59.56 -> 67.52 | Improved |
| apply occlusion filter at evaluation | CenterTrack · PersonPath22 test | MOTA: 52.31 -> 59.28 | Improved |
| apply occlusion filter at evaluation | FairMOT · PersonPath22 test | MOTA: 54.29 -> 61.79 | Improved |
| apply occlusion filter at evaluation | FairMOT · PersonPath22 test | IDF1: 56.52 -> 61.05 | Improved |
| apply occlusion filter at evaluation | CenterTrack · PersonPath22 test | IDF1: 43.04 -> 46.36 | Improved |
| use Faster R-CNN detector instead of CenterNet | tracking detector · PersonPath22 test | AP@0.5: 78.51% -> 82.03% | Improved |
| use person embedding re-identification (FairMOT) instead of motion-based linking (SiamMOT) | online tracker · PersonPath22 test (occlusion filter) | IDF1: 53.71 -> 61.05 | Improved |
| track large-size instead of small-size persons | FairMOT · PersonPath22 test, size subsets | MOTA: 40.8 -> 72.0 | Improved |
| track static/slow instead of moving persons | CenterTrack · PersonPath22 test, motion subsets | MOTA: 43.7 -> 57.1 | Improved |
| track static/slow instead of moving persons | SiamMOT · PersonPath22 test, motion subsets | IDF1: 52.6 -> 52.6 | No clear effect |
| track short-to-no occlusion instead of medium-to-long occlusion tracks | FairMOT · PersonPath22 test, occlusion subsets | IDF1: 44.3 -> 68.3 | Improved |
| long instead of short track duration | FairMOT · PersonPath22 test, duration subsets | IDF1: 68.3 -> 44.3 | Hurt |
Closest papers by meaning
- PoseTrack21: A Dataset for Person Search, Multi-Object Tracking and Multi-Person Pose Tracking (CVPR 2022 · similarity 0.88)
- MTMMC: A Large-Scale Real-World Multi-Modal Camera Tracking Benchmark (CVPR 2024 · similarity 0.88)
- TAO: A Large-Scale Benchmark for Tracking Any Object (ECCV 2020 · similarity 0.88)
- MOTChallenge: A Benchmark for Single-Camera Multiple Target Tracking (IJCV 2020 · similarity 0.87)
- Motions as Queries: One-Stage Multi-Person Holistic Human Motion Capture (CVPR 2025 · similarity 0.86)
- Delving into the Trajectory Long-tail Distribution for Muti-object Tracking (CVPR 2024 · similarity 0.85)
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.
