Research map / Detection and segmentation / Tracking
Reading Relevant Feature from Global Representation Memory for Visual Object 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 fixed-interval template replacement with GR memory token selection (ablation) | RFGM · LaSOT | AUC: 64.5 -> 70.3 | Helped |
| replace fixed-interval template replacement with GR memory token selection (ablation) | RFGM · TrackingNet | AUC: 84.5 -> 84.7 | Helped |
| replace fixed-interval template replacement with GR memory token selection (ablation) | RFGM · GOT-10k | AO: 71.9 -> 74.1 | Mixed |
| use prediction-head center-score template updates instead of GR memory (ablation) | RFGM · GOT-10k | AO: 74.3 vs 74.1 | No clear effect |
| replace vanilla ViT attention with relevance attention (ablation) | RFGM with GR memory · TrackingNet | AUC: 84.5 -> 84.7 | Helped |
| replace vanilla ViT attention with relevance attention (ablation) | RFGM with GR memory · GOT-10k | AO: 73.9 -> 74.1 | Helped |
| replace vanilla ViT attention with relevance attention (ablation) | RFGM with GR memory · LaSOT | AUC: 69.8 -> 70.3 | Helped |
| increase GR memory size from 64 to 192 tokens (ablation) | RFGM · LaSOT | AUC: 69.0 -> 70.3 | Helped |
| increase GR memory size from 192 to 256 tokens (ablation) | RFGM · LaSOT | AUC: 70.3 -> 69.6 | Hurt |
| increase GR memory size from 192 to 448 tokens (ablation) | RFGM · LaSOT | AUC: 70.3 -> 69.5 | Hurt |
| use relevance attention at 3 stages instead of no relevance-attention stages (ablation) | RFGM · model computation | MACs: 40.00G -> 37.66G | Helped |
| use relevance attention instead of vanilla attention (ablation) | RFGM · TrackingNet | P: 83.7 -> 83.6 | No clear effect |
Closest papers by meaning
- A Distractor-Aware Memory for Visual Object Tracking with SAM2 (CVPR 2025 · similarity 0.88)
- Less Is More: Token Context-Aware Learning for Object Tracking (AAAI 2025 · similarity 0.86)
- Exploring Enhanced Contextual Information for Video-Level Object Tracking (AAAI 2025 · similarity 0.86)
- ODTrack: Online Dense Temporal Token Learning for Visual Tracking (AAAI 2024 · similarity 0.85)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.85)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 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.