Research map / Reinforcement learning / Policy gradient and actor-critic
Real-time 'Actor-Critic' Tracking
Boyu Chen, Dong Wang, Peixia Li, Shuang Wang, Huchuan Lu
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 reinforcement learning (Critic-guided DDPG) offline training (ablation) | ACT · OTB-2013 | precision (20px): 0.876 -> 0.905 | Improved |
| add reinforcement learning (Critic-guided DDPG) offline training (ablation) | ACT · OTB-2013 | success AUC: 0.639 -> 0.657 | Improved |
| use expert decision guidance instead of normal DDPG exploration (ablation) | ACT · OTB-2013 | precision (20px): 0.861 -> 0.905 | Improved |
| use expert decision guidance instead of normal DDPG exploration (ablation) | ACT · OTB-2013 | success AUC: 0.616 -> 0.657 | Improved |
| add first-frame supervised initialization of Actor during offline training (ablation) | ACT · OTB-2013 | precision (20px): 0.849 -> 0.905 | Improved |
| add first-frame supervised initialization of Actor during offline training (ablation) | ACT · OTB-2013 | success AUC: 0.577 -> 0.657 | Improved |
| add offline pretraining of Actor rather than plain VGG-M init (ablation) | ACT · OTB-2013 | precision (20px): 0.787 -> 0.905 | Improved |
| add offline pretraining of Actor rather than plain VGG-M init (ablation) | ACT · OTB-2013 | success AUC: 0.550 -> 0.657 | Improved |
| use one continuous action (ACT) instead of iterative discrete actions (ADNet) | ACT · OTB-2013 | FPS: 3 -> 30 | Improved |
| use one continuous action (ACT) instead of iterative discrete actions (ADNet) | ACT · OTB-2015 | precision (20px): 0.880 -> 0.859 | Hurt |
| use ACT instead of MDNet random sampling search | ACT · OTB-2013 | precision (20px): 0.948 -> 0.905 | Hurt |
Closest papers by meaning
- Deep Reinforcement Learning with Iterative Shift for Visual Tracking (ECCV 2018 · similarity 0.90)
- Visual Tracking via Hierarchical Deep Reinforcement Learning (AAAI 2021 · similarity 0.89)
- Tracking as Online Decision-Making: Learning a Policy From Streaming Videos With Reinforcement Learning (ICCV 2017 · similarity 0.88)
- Action-Decision Networks for Visual Tracking With Deep Reinforcement Learning (CVPR 2017 · similarity 0.88)
- Online Decision Based Visual Tracking via Reinforcement Learning (NeurIPS 2020 · similarity 0.88)
- POST: POlicy-Based Switch Tracking (AAAI 2020 · 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.
