Research map / Detection and segmentation / Tracking
"FEAR: Fast, Efficient, Accurate and Robust Visual Tracker"
Vasyl Borsuk, Roman Vei, Orest Kupyn, Tetiana Martyniuk, Igor Krashenyi, Jiři Matas
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 lower backbone spatial resolution (original stride) (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | iPhone 11 FPS: 122.19 -> 208.41 | Improved |
| add lower backbone spatial resolution (original stride) (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | EAO: 0.236 -> 0.234 | No clear effect |
| add pixel-wise fusion block (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | EAO: 0.234 -> 0.264 | Improved |
| add pixel-wise fusion block (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | iPhone 11 FPS: 208.41 -> 207.72 | No clear effect |
| add dynamic template update (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | EAO: 0.264 -> 0.270 | Improved |
| add dynamic template update (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | Robustness: 0.683 -> 0.708 | Improved |
| add dynamic template update (ablation) | FEAR-XS · VOT-ST2021 (FEAR-XS ablation) | iPhone 11 FPS: 207.72 -> 205.12 | No clear effect |
| evaluate tracker on 240 FPS frames instead of 30 FPS (ablation) | FEAR-XS · NFS | Success Score: 0.618 -> 0.655 | Improved |
| evaluate tracker on 240 FPS frames instead of 30 FPS (ablation) | FEAR-XS · NFS | Precision Score: 0.753 -> 0.816 | Improved |
| run models in float16 instead of float32 (ablation) | FEAR trackers · mobile inference | accuracy: ±0.5% | No clear effect |
| keep original backbone resolution rather than upscaling (ablation) | FEAR-XS · FLOPs | backbone GigaFLOPs: 0.840 -> 0.318 | Improved |
Closest papers by meaning
- SiamFC++: Towards Robust and Accurate Visual Tracking with Target Estimation Guidelines (AAAI 2020 · similarity 0.89)
- Visual Object Tracking with Discriminative Filters and Siamese Networks: A Survey and Outlook (TPAMI 2022 · similarity 0.87)
- Ranking-Based Siamese Visual Tracking (CVPR 2022 · similarity 0.87)
- Two-stream Beats One-stream: Asymmetric Siamese Network for Efficient Visual Tracking (AAAI 2025 · similarity 0.87)
- STMTrack: Template-Free Visual Tracking With Space-Time Memory Networks (CVPR 2021 · similarity 0.87)
- SiamMOT: Siamese Multi-Object Tracking (CVPR 2021 · 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.
