Research map / Video / Action recognition
Part-Activated Deep Reinforcement Learning for Action Prediction
Lei Chen, Jiwen Lu, Zhanjie Song, Jie Zhou
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-learned part activation (ablation) | skeleton proposal features · UTI Set #1 (OR=0.5) | accuracy (%): 76.7 -> 91.7 | Improved |
| add reinforcement-learned part activation (ablation) | skeleton proposal features · UTI Set #1 (OR=1.0) | accuracy (%): 91.7 -> 96.7 | Improved |
| add reinforcement-learned part activation (ablation) | skeleton proposal features · UTI Set #2 (OR=0.5) | accuracy (%): 70.0 -> 83.3 | Improved |
| add reinforcement-learned part activation (ablation) | skeleton proposal features · UTI Set #2 (OR=1.0) | accuracy (%): 86.7 -> 91.7 | Improved |
| add reinforcement-learned part activation (ablation) | skeleton proposal features · BIT-Interaction (OR=0.5) | accuracy (%): 68.6 -> 85.9 | Improved |
| add reinforcement-learned part activation (ablation) | skeleton proposal features · BIT-Interaction (OR=1.0) | accuracy (%): 87.5 -> 91.4 | Improved |
| add local patch appearance features around joints to skeleton-only features (ablation) | skeleton feature · UTI Set #1 (OR=0.5) | accuracy (%): 69.9 -> 76.7 | Improved |
| add local patch appearance features around joints to skeleton-only features (ablation) | skeleton feature · BIT-Interaction (OR=1.0) | accuracy (%): 75.7 -> 87.5 | Improved |
| add local patch appearance features around joints to skeleton-only features (ablation) | skeleton feature · UTI Set #2 (OR=0.5) | accuracy (%): 66.7 -> 70.0 | Improved |
| use PA-DRL instead of AAC at half observation | AAC · UTI Set #1 (OR=0.5) | accuracy (%): 88.3 -> 91.7 | Improved |
| use PA-DRL instead of C3D recognition model at half observation | C3D · UCF101 (OR=0.5) | accuracy (%): 80.0 -> 87.3 | Improved |
| use PA-DRL instead of AAC at full observation on UTI Set #2 | AAC · UTI Set #2 (OR=1.0) | accuracy (%): 63.9 -> 91.7 | Improved |
Closest papers by meaning
- Deep Progressive Reinforcement Learning for Skeleton-Based Action Recognition (CVPR 2018 · similarity 0.89)
- Imitation Learning for Human Pose Prediction (ICCV 2019 · similarity 0.85)
- Deep Reinforcement Learning for Active Human Pose Estimation (AAAI 2020 · similarity 0.84)
- Whole-Body Conditioned Egocentric Video Prediction (NeurIPS 2025 · similarity 0.83)
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.83)
- Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes (AAAI 2025 · similarity 0.82)
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.
