Research map / Video / Video generation
LDTrack: Dynamic People Tracking by Service Robots Using Diffusion Models
Angus Fung, Beno Benhabib, Goldie Nejat
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 |
|---|---|---|---|
| increase number of past track embeddings from 1 to 5 (ablation) | LDTrack · IOD | MOTA: 79.7 -> 79.2 | No clear effect |
| increase number of past track embeddings from 1 to 5 (ablation) | LDTrack · R-CPC Test Set 1 | MOTA: 49.0 -> 48.8 | No clear effect |
| raise person box embeddings from 250 to 500 (ablation) | LDTrack · R-CPC Test Set 1 | MOTA: 47.2 -> 49.0 | Improved |
| raise person box embeddings from 500 to 1000 (ablation) | LDTrack · R-CPC Test Set 1 | MOTA: 49.0 -> 49.0 | No clear effect |
| raise latent dimension from 144 to 288 (ablation) | LDTrack · R-CPC Test Set 1 | MOTA: 46.6 -> 49.0 | Improved |
| raise latent dimension from 288 to 566 (ablation) | LDTrack · R-CPC Test Set 1 | MOTA: 49.0 -> 49.3 | No clear effect |
| raise latent dimension from 144 to 288 (ablation) | LDTrack · IOD | MOTA: 76.7 -> 79.7 | Improved |
| track-conditioned latent diffusion versus SORT | LDTrack · IOD | MOTA: 57.3 -> 78.6 | Improved |
| LDTrack versus REGROUP | people tracking · R-CPC Test Set 1 | MOTA: 38.4 -> 49.0 | Improved |
| RGB-only LDTrack versus RGB-D methods | KTP · KTP (occlusion) | MOTA: 88.2 -> 92.7 | Improved |
| LDTrack versus DiffusionTrack (single frame input) | MOT17 · MOT17 test | MOTA: 77.9 -> 76.3 | Hurt |
| LDTrack versus DiffusionTrack | MOT20 · MOT20 test | MOTA: 72.8 -> 68.9 | Hurt |
Closest papers by meaning
- Multi-view Crowd Tracking Transformer with View-Ground Interactions Under Large Real-World Scenes (CVPR 2026 · similarity 0.83)
- PoseTrack21: A Dataset for Person Search, Multi-Object Tracking and Multi-Person Pose Tracking (CVPR 2022 · similarity 0.83)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.83)
- MotionTrack: Learning Robust Short-Term and Long-Term Motions for Multi-Object Tracking (CVPR 2023 · similarity 0.83)
- DiffusionTrack: Point Set Diffusion Model for Visual Object Tracking (CVPR 2024 · similarity 0.83)
- DINTR: Tracking via Diffusion-based Interpolation (NeurIPS 2024 · similarity 0.82)
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.
