Research map / Label-efficient and robust learning / Semi-supervised learning
Unsupervised Deep Tracking
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 |
|---|---|---|---|
| add adaptive online model update and improved DCF formulation (ablation) | UDT · OTB-2015 | AUC score (%): 59.4 -> 63.2 | Helped |
| add adaptive online model update and improved DCF formulation (ablation) | UDT · VOT2016 | EAO: 0.226 -> 0.301 | Helped |
| train with fully supervised labels instead of unlabeled video (ablation) | UDT · OTB-2015 | AUC: 3% | Helped |
| add multiple-frame validation and cost-sensitive loss (ablation) | UDT · OTB-2015 | tracking accuracy | Helped |
| add motion-based hard-sample reweighting (ablation) | UDT · OTB-2015 | tracking accuracy | Helped |
| add multi-frame forward-backward validation instead of single-trajectory validation (ablation) | UDT · OTB-2015 | tracking accuracy | Helped |
| use weakly labeled training samples rather than randomly cropped unlabeled patches (ablation) | UDT · OTB-2015 | tracking accuracy | No clear effect |
| fine-tune using the first 5 frames and first-frame bounding-box annotation (ablation) | UDT · OTB-2015 | tracking accuracy | Helped |
| add 337 unlabeled OxUvA videos to training (ablation) | UDT · OTB-2015 | distance precision (DP): +0.9% DP | Helped |
| add 337 unlabeled OxUvA videos to training (ablation) | UDT · OTB-2015 | AUC: +0.7% AUC | Helped |
| use unsupervised rather than fully supervised training (ablation) | UDT · OTB-2015 | tracking accuracy: 3% | Hurt |
| use unsupervised rather than fully supervised training (ablation) | UDT · OTB-2015 attributes: illumination variation, occlusion, and fast motion | tracking accuracy | Hurt |
Closest papers by meaning
- Learning a Visual Tracker from a Single Movie without Annotation (AAAI 2019 · similarity 0.93)
- Unsupervised Learning of Accurate Siamese Tracking (CVPR 2022 · similarity 0.92)
- Progressive Unsupervised Learning for Visual Object Tracking (CVPR 2021 · similarity 0.91)
- Correlation-Aware Deep Tracking (CVPR 2022 · similarity 0.89)
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision (NeurIPS 2025 · similarity 0.86)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.86)
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.