Research map / Detection and segmentation / Tracking
Beyond Local Search: Tracking Objects Everywhere With Instance-Specific Proposals
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| replace Struck's local search with instance-specific proposals across the frame | Struck / EBT · OTB OPE | AUC: 47.2 -> 58.1 | Helped |
| replace Struck's local search with instance-specific proposals across the frame | Struck / EBT · TB50 OPE | AUC: 36.3 -> 49.6 | Helped |
| use instance-specific proposals with global search | Struck / EBT · VOT2014 | Final Rank: 22.83 -> 13.03 | Helped |
| use instance-specific proposals with global search | Struck / EBT · OTB Fast Motion (17) | AUC: 45.7 -> 58.1 | Helped |
| use instance-specific proposals with global search | Struck / EBT · TB50 Fast Motion (25) | AUC: 34.4 -> 53.3 | Helped |
| use instance-specific proposals with global search | Struck / EBT · VOT2014+ | AUC: 28.4 -> 43.7 | Helped |
| use 200 rather than 100 proposals (ablation) | EBT · TB50 OPE | AUC: 46.8 -> 49.6 | Helped |
| increase the proposal count from 200 to 400 (ablation) | EBT · TB50 OPE | AUC: 49.6 -> 48.0 | Hurt |
| use instance-specific re-ranking rather than raw EdgeBox proposals (ablation) | EBT · TB50 OPE | AUC: 47.3 -> 49.6 | Helped |
| add a motion smoothness term (ablation) | EBT · TB50 OPE | AUC: 46.2 -> 49.6 | Helped |
| use 2640-D rather than 480-D features (ablation) | EBT · TB50 OPE | AUC: 45.5 -> 49.6 | Helped |
| use global proposals only for testing and combine proposals with local samples for model updates (ablation) | EBT · TB50 OPE | AUC: 39.2/56.5 -> 49.6/73.9 | Helped |
Closest papers by meaning
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.90)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.89)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.89)
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision (NeurIPS 2025 · similarity 0.88)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.87)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 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.