Research map / Detection and segmentation / Tracking
Joint Feature Learning and Relation Modeling for Tracking: A One-Stream Framework
Botao Ye, Hong Chang, Bingpeng Ma, Shiguang Shan, Xilin Chen
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 early candidate elimination (ablation) | OSTrack-256 · LaSOT | AUC: 68.7 -> 69.1 | Improved |
| add early candidate elimination (ablation) | OSTrack-256 · LaSOT | FPS: 93.1 -> 105.4 | Improved |
| add early candidate elimination (ablation) | OSTrack-384 · LaSOT | FPS: 41.4 -> 58.1 | Improved |
| add early candidate elimination (ablation) | OSTrack-256 · LaSOT | MACs (G): 29.0 -> 21.5 | Improved |
| use MAE instead of ImageNet-21k pre-training (ablation) | OSTrack-256 · LaSOT | AUC: 66.9 -> 68.7 | Improved |
| add pre-training (ablation) | OSTrack-256 · LaSOT | AUC: 60.4 -> 68.7 | Improved |
| use one-stream structure instead of STARK-aligned two-stream (ablation) | OSTrack-256 · GOT-10k | AO: 68.8 -> 71.0 | Improved |
| use one-stream structure instead of SwinTrack-aligned two-stream (ablation) | OSTrack-256 · Aligned comparison | FPS: 67.5 -> 93.1 | Improved |
| use all template tokens for elimination instead of baseline without elimination (ablation) | OSTrack-256 · LaSOT | AUC: 68.7 -> 68.1 | Hurt |
| use center template token instead of all template tokens for elimination (ablation) | OSTrack-256 · LaSOT | AUC: 68.1 -> 69.1 | Improved |
| add identity embeddings (ablation) | OSTrack-256 · LaSOT | AUC: 68.7 -> 68.0 | Hurt |
| add relative positional embeddings (ablation) | OSTrack-256 · LaSOT | AUC: 68.7 -> 68.5 | No clear effect |
Closest papers by meaning
- OneTrack: Demystifying the Conflict Between Detection and Tracking in End-to-End 3D Trackers (ECCV 2024 · similarity 0.86)
- SOVTrack: Open-Vocabulary Multi-Object Tracking with Self-Supervised Pseudo Labeling and Feature Distillation (ECCV 2026 · similarity 0.86)
- OneTracker: Unifying Visual Object Tracking with Foundation Models and Efficient Tuning (CVPR 2024 · similarity 0.84)
- SUTrack: Towards Simple and Unified Single Object Tracking (AAAI 2025 · similarity 0.84)
- SPMTrack: Spatio-Temporal Parameter-Efficient Fine-Tuning with Mixture of Experts for Scalable Visual Tracking (CVPR 2025 · similarity 0.84)
- Two-stream Beats One-stream: Asymmetric Siamese Network for Efficient Visual Tracking (AAAI 2025 · similarity 0.83)
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.
