Research map / Detection and segmentation / Tracking
Quo Vadis: Is Trajectory Forecasting the Key Towards Long-Term Multi-Object Tracking?
Patrick Dendorfer, Vladimir Yugay, Aljosa Osep, Laura Leal-Taixé
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 BEV trajectory forecasting to ByteTrack | ByteTrack · MOT17 test | HOTA: 63.05 -> 63.14 | Improved |
| Add BEV trajectory forecasting to ByteTrack | ByteTrack · MOT20 test | HOTA: 61.34 -> 61.48 | Improved |
| Add BEV trajectory forecasting and egomotion estimation | CenterTrack · MOT17 validation, moving scenes | HOTA: +3.07 | Improved |
| Use BEV coordinates rather than pixel coordinates for linear forecasting (ablation) | QuoVadis · MOTSynth occlusion-gap matching | successful matches: 50.2% -> 89.3% | Improved |
| Use data-driven homography for BEV localization rather than per-frame monocular depth (ablation) | QuoVadis · MOTSynth occlusion-gap matching | successful matches: 43.1% -> 74.3% | Improved |
| Add a social interaction module to a 3-sample GAN (ablation) | GAN trajectory forecaster · MOT17 validation | HOTA: 54.43 -> 54.52 | No clear effect |
| Increase vanilla GAN forecast samples from 3 to 20 (ablation) | GAN trajectory forecaster · MOT17 validation | HOTA: 54.43 -> 53.81 | Hurt |
| Use multimodal generator heads to create diverse forecasts (ablation) | GAN trajectory forecaster · MOT17 validation | ID lost, long occlusions: -8.64% -> -17.43% | Improved |
| Use L2 plus IoU matching without thresholds (ablation) | QuoVadis · MOT17 validation | AssRe: 61.08 -> 61.17 | Improved |
| Use L2 plus IoU matching without thresholds (ablation) | QuoVadis · MOT17 validation | AssPr: 72.36 -> 69.27 | Hurt |
| Add appearance and IoU thresholds to L2-plus-IoU matching (ablation) | QuoVadis · MOT17 validation | HOTA: 53.75 -> 54.27 | Improved |
| Add forecasting to existing trackers | Top-ranked MOT trackers · MOT17 validation, static scenes | IDSW: -3 to -146 | Improved |
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 2 of 2 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
- Standing Between Past and Future: Spatio-Temporal Modeling for Multi-Camera 3D Multi-Object Tracking (CVPR 2023 · similarity 0.89)
- Object Tracking using Spatio-Temporal Networks for Future Prediction Location (ECCV 2020 · similarity 0.88)
- OVTrack: Open-Vocabulary Multiple Object Tracking (CVPR 2023 · similarity 0.87)
- S2-Track: A Simple yet Strong Approach for End-to-End 3D Multi-Object Tracking (ICML 2025 · similarity 0.86)
- Samba: Synchronized Set-of-Sequences Modeling for Multiple Object Tracking (ICLR 2025 · similarity 0.86)
- TGFormer: Transformer with Track Query Group for Multi-Object Tracking (AAAI 2025 · similarity 0.86)
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.
