Research map / Detection and segmentation / Tracking
Tracking by Instance Detection: A Meta-Learning Approach
Guangting Wang, Chong Luo, Xiaoyan Sun, Zhiwei Xiong, Wenjun Zeng
Check 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 |
|---|---|---|---|
| Use MAML initialization rather than standard detector training, then adapt on the initial frame (ablation) | FCOS detector · OTB-100 | AUC: 0.464 -> 0.671 | Improved |
| Add kernel-wise learnable learning rates to both classification and regression branches (ablation) | FCOS-MAML · OTB-100 | AUC: 0.628 -> 0.704 | Improved |
| Add KLLR to the classification branch (ablation) | FCOS-MAML · OTB-100 | AUC: 0.628 -> 0.661 | Improved |
| Add KLLR to the regression branch (ablation) | FCOS-MAML · OTB-100 | AUC: 0.628 -> 0.676 | Improved |
| Online-update both classification and regression branches instead of only classification (ablation) | FCOS-MAML · LaSOT | AUC: 0.523 -> 0.496 | Hurt |
| Online-update both classification and regression branches instead of only classification (ablation) | FCOS-MAML · VOT-2018 | EAO: 0.394 -> 0.392 | No clear effect |
| Online-update the classification branch rather than omitting online updating (ablation) | FCOS-MAML · OTB-100 | AUC: 0.671 -> 0.690 | Improved |
| Share block-4 between classification and regression branches (ablation) | FCOS-MAML · OTB-100 | AUC: 0.704 -> 0.712 | No clear effect |
| Replace ResNet-18 with ResNet-50 (ablation) | FCOS-MAML · VOT-2018 | EAO: 0.392 -> 0.444 | Improved |
| Replace ResNet-18 with ResNet-50 (ablation) | FCOS-MAML · TrackingNet | AUC: 0.757 -> 0.758 | No clear effect |
| Use FCOS-MAML instead of DiMP-18 | TrackingNet · TrackingNet | AUC: 0.723 -> 0.757 | Improved |
| Use FCOS-MAML instead of DiMP-18 | LaSOT-test · LaSOT-test | AUC: 0.532 -> 0.523 | Hurt |
Closest papers by meaning
- Meta-Tracker: Fast and Robust Online Adaptation for Visual Object Trackers (ECCV 2018 · similarity 0.91)
- RetinaTrack: Online Single Stage Joint Detection and Tracking (CVPR 2020 · similarity 0.89)
- Real-Time Object Tracking via Meta-Learning: Efficient Model Adaptation and One-Shot Channel Pruning (AAAI 2020 · similarity 0.89)
- Bridging the Gap Between Detection and Tracking: A Unified Approach (ICCV 2019 · similarity 0.88)
- OmniTracker: Unifying Visual Object Tracking by Tracking-With-Detection (TPAMI 2025 · similarity 0.88)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 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.
