Research map / Detection and segmentation / Tracking
Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths
Andrea Hornakova, Timo Kaiser, Paul Swoboda, Michal Rolinek, Bodo Rosenhahn, Roberto Henschel
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 message passing (82 steps) on dense lifted graph with zero base cost (ablation) | ApLift · MOT17 train | IDF1: 70.3 -> 71.0 | Improved |
| use dense instead of sparsified lifted graph (ablation) | ApLift · MOT17 train | IDF1: 69.1 -> 69.8 | Improved |
| set base edge costs to zero instead of keeping original base costs (ablation) | ApLift (dense) · MOT17 train | IDF1: 69.8 -> 71.0 | Improved |
| add message passing (82 steps) (ablation) | ApLift · MOT17 train | MOTA: 66.3 -> 66.3 | No clear effect |
| use approximate solver (6 iterations) instead of two-step LifT optimal solver | LDP tracker · MOT20-01 (200 frames) | IDF1: 73.2 -> 75.4 | Improved |
| use approximate solver (6 iterations) instead of two-step LifT optimal solver | LDP tracker · MOT20-01 (200 frames) | runtime (s): 2807 -> 36 | Improved |
| run more message passing iterations (51 instead of 6) (ablation) | ApLift · MOT20-01 (200 frames) | IDF1: 75.4 -> 74.6 (time 36 s -> 616 s) | Hurt |
| use approximate solver instead of optimal LDP solver LifT | LDP tracker · MOT16 test | MOTA: 61.3 -> 61.7 (IDF1 64.7 -> 66.1) | Improved |
| use approximate solver instead of optimal LDP solver LifT | LDP tracker · MOT17 test | MOTA: 60.5 -> 60.5 (IDF1 65.6 -> 65.6) | No clear effect |
| use approximate solver instead of optimal LDP solver LifT | LDP tracker · MOT15 test | MOTA: 52.5 -> 51.1 (IDF1 60.0 -> 59.0) | Hurt |
Closest papers by meaning
- Lifted Disjoint Paths with Application in Multiple Object Tracking (ICML 2020 · similarity 0.89)
- MOTCOM: The Multi-Object Tracking Dataset Complexity Metric (ECCV 2022 · similarity 0.87)
- Sampling-Resilient Multi-Object Tracking (AAAI 2024 · similarity 0.85)
- DiffMOT: A Real-time Diffusion-based Multiple Object Tracker with Non-linear Prediction (CVPR 2024 · similarity 0.85)
- S2-Track: A Simple yet Strong Approach for End-to-End 3D Multi-Object Tracking (ICML 2025 · similarity 0.85)
- Towards Generalizable Multi-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.
