Research map / Detection and segmentation / Tracking
GlobalTrack: A Simple and Strong Baseline for Long-Term Tracking
Read the paperCheck 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 local search with full-image global instance search | GlobalTrack · long-term tracking, including TLP | SR@0.5: 52.7 -> 63.8 | Helped |
| train with cross-query loss instead of single-query loss (ablation) | GlobalTrack · LaSOT test | Success: 49.5 -> 52.1 | Helped |
| train with cross-query loss instead of single-query loss (ablation) | GlobalTrack · LaSOT test | Precision: 49.3 -> 52.7 | Helped |
| train with cross-query loss instead of single-query loss (ablation) | GlobalTrack · LaSOT test | Norm. Precision: 55.7 -> 59.9 | Helped |
| use query-guided proposals rather than generic RPN proposals | RPN proposal generation · RPN proposal evaluation | AR@1: 11.3 -> 67.1 | Helped |
| use query-guided proposals rather than GA-RPN proposals | RPN proposal generation · RPN proposal evaluation | AR@1: 12.9 -> 67.1 | Helped |
| use query-guided RPN rather than query-guided RCNN for top-1 prediction | GlobalTrack · proposal evaluation | AR@1: 67.1 -> 76.6 | Hurt |
| use query-guided RPN rather than query-guided RCNN for multi-candidate recall | GlobalTrack · proposal evaluation | AR@5: 82.5 -> 87.1 | Helped |
| compare GlobalTrack with SiamRPN++ on TrackingNet | GlobalTrack · TrackingNet test | Success: 73.3 -> 70.4 | Hurt |
| compare GlobalTrack with SPLT on OxUvA test | GlobalTrack · OxUvA test | MaxGM: 62.2 -> 60.3 | Hurt |
Closest papers by meaning
- Global Tracking via Ensemble of Local Trackers (CVPR 2022 · similarity 0.92)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.90)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.89)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.87)
- SUTrack: Towards Simple and Unified Single Object Tracking (AAAI 2025 · similarity 0.87)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · 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.