Research map / Detection and segmentation / Tracking
Towards Frame Rate Agnostic Multi-object Tracking
Weitao Feng, Lei Bai, Yongqiang Yao, Fengwei Yu, Wanli Ouyang
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 |
|---|---|---|---|
| add PTS training (ablation) | unified model baseline · MOT20 val (unknown frame rate) | mHOTA: 56.5 -> 58.6 | Improved |
| add PTS training (ablation) | unified model baseline · MOT20 val (unknown frame rate) | VR: 34.6 -> 26.7 | Improved |
| add FAAM module (ablation) | unified model baseline · MOT20 val (unknown frame rate) | mHOTA: 56.5 -> 59.1 | Improved |
| add PTS training on top of FAAM (ablation) | FAAM · MOT20 val (unknown frame rate) | mHOTA: 59.1 -> 61.0 | Improved |
| add PTS training on top of FAAM (ablation) | FAAM · MOT20 val (known frame rate) | VR: 28.5 -> 24.8 | Improved |
| use FAAM+PTS (1x params) instead of three frame-rate-specific models (ablation) | Baseline (MM) · MOT20 val (known frame rate) | mHOTA: 58.9 -> 61.1 | Improved |
| use FAAM+PTS instead of baselines | Baseline (MM) · MOT20 val, unseen frame rates | mHOTA: 57.0 -> 59.6 | Improved |
| use FAAM+PTS instead of baselines | Baseline (UM) · MOT20 val, dynamic sampling | HOTA: 50.5 -> 55.5 | Improved |
| use FAAM+PTS under fast object motion | Baseline (UM) · MOT17 val, fast motion (19 ppf) | mHOTA: 54.2 -> 53.8 | No clear effect |
| use FAAM+PTS under normal motion | Baseline (UM) · MOT17 val, normal motion (8-14 ppf) | mHOTA: 50.7 -> 54.5 | Improved |
| add period number Np=3 in PTS (ablation) | FAAM+PTS · MOT20 val | mHOTA: 59.1 -> 61.0 | Improved |
| vary matching threshold from 0.5 to 1.3 (ablation) | FAAM+PTS · MOT20 val | mHOTA: 60.8 -> 61.1 | No clear effect |
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.
- ByteTrack — ByteTrack: Multi-Object Tracking by Associating Every Detection Box (ECCV 2022) · better on 9 of 9 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
- Need for Speed: A Benchmark for Higher Frame Rate Object Tracking (ICCV 2017 · similarity 0.89)
- Sampling-Resilient Multi-Object Tracking (AAAI 2024 · similarity 0.88)
- MOTS: Multi-Object Tracking and Segmentation (CVPR 2019 · similarity 0.88)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.87)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.87)
- More Than Meets the Eye: Enhancing Multi-Object Tracking Even with Prolonged Occlusions (ICML 2025 · similarity 0.87)
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.
