Research map / Detection and segmentation / Tracking
Real-Time Object Tracking via Meta-Learning: Efficient Model Adaptation and One-Shot Channel Pruning
Ilchae Jung, Kihyun You, Hyeonwoo Noh, Minsu Cho, Bohyung Han
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 |
|---|---|---|---|
| Use meta-learned adaptation instead of standard RT-MDNet adaptation | RT-MDNet · OTB2015 | success (%): 65.0 -> 65.5 | Improved |
| Use meta-learned adaptation with 5 initial and 5 online iterations | RT-MDNet · OTB2015 | FPS: 42 -> 58 | Improved |
| Add estimated-target online adaptation (ablation) | MetaRTT · OTB2015 | success (%): 58.6 -> 65.5 | Improved |
| Add hard examples to meta-training (ablation) | MetaRTT · OTB2015 | success (%): 60.7 -> 65.5 | Improved |
| Use per-parameter and per-iteration learning rates instead of a scalar rate (ablation) | MetaRTT · OTB2015 | success (%): 60.9 -> 65.5 | Improved |
| Add one-shot channel pruning | MetaRTT · OTB2015 | FPS: 58 -> 65 | Improved |
| Add one-shot channel pruning | MetaRTT · OTB2015 | success (%): 65.5 -> 65.1 | Hurt |
| Meta-learn the pruning network over complete simulated episodes (ablation) | MetaRTT+Prune · OTB2015 | success (%): 61.7 -> 65.1 | Improved |
| Use input-adaptive pruning instead of a universal static mask (ablation) | MetaRTT+Prune · OTB2015 | success (%): 60.9 -> 65.1 | Improved |
| Reduce RT-MDNet adaptation to 5 initial and 5 online iterations without meta-learning (ablation) | RT-MDNet · OTB2015 | success (%): 65.0 -> 57.8 | Hurt |
| Add COCO pretraining | MetaRTT · VOT2016 | EAO: 0.346 -> 0.350 | Improved |
Closest papers by meaning
- Meta-Tracker: Fast and Robust Online Adaptation for Visual Object Trackers (ECCV 2018 · similarity 0.92)
- Deep Meta Learning for Real-Time Target-Aware Visual Tracking (ICCV 2019 · similarity 0.90)
- Tracking by Instance Detection: A Meta-Learning Approach (CVPR 2020 · similarity 0.89)
- Tracking by Animation: Unsupervised Learning of Multi-Object Attentive Trackers (CVPR 2019 · similarity 0.86)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.85)
- Sampling-Resilient Multi-Object Tracking (AAAI 2024 · similarity 0.84)
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.
