Research map / 3D vision / Human pose estimation
Mining 3D Key-Pose-Motifs for Action Recognition
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 |
|---|---|---|---|
| Mine key-pose-motifs from activated-simplex soft-quantized pose sequences | 3D skeletal action recognition · MSR-Action3D, all 252 splits | accuracy: 89.00 -> 94.40 | Helped |
| Mine key-pose-motifs from activated-simplex soft-quantized pose sequences | 3D skeletal action recognition · UTKinect, leave-one-sequence-out | accuracy: 91.5 -> 93.47 | Helped |
| Use skeleton-only key-pose-motif recognition | 3D skeletal action recognition · MSR Daily Activity3D, single split | accuracy: 80.00 -> 83.47 | Helped |
| Use skeleton-only key-pose-motif recognition | 3D skeletal action recognition · Florence, leave-one-actor-out | accuracy: 90.88 -> 92.25 | Helped |
| Use key-pose-motifs rather than direct sequence matching | MSR-Action3D · MSR-Action3D | accuracy: 88% -> 94.40 | Helped |
| Add soft probabilistic quantization instead of assigning one symbol per action-unit (ablation) | Key-pose-motif mining · MSR-Action3D | accuracy: 91% -> 94.40 | Helped |
| Use activated simplices instead of k-means for probabilistic quantization (ablation) | Key-pose-motif mining · MSR-Action3D | accuracy: 88.7% -> 94.40 | Helped |
| Allow gaps between consecutive poses in motifs (ablation) | Key-pose-motif recognition · MSR-Action3D | accuracy | Helped |
| Increase the maximum motif gap beyond 20 (ablation) | Key-pose-motif recognition · MSR-Action3D | accuracy | Hurt |
| Increase the number of mined motifs when it is below 80 (ablation) | Key-pose-motif recognition · MSR-Action3D | accuracy | Helped |
| Mine too many motifs, including motifs present in only a few sequences (ablation) | Key-pose-motif recognition · MSR-Action3D | accuracy | Hurt |
| Combine more consecutive poses into each action-unit (ablation) | Key-pose-motif recognition · MSR-Action3D | accuracy | Helped |
Closest papers by meaning
- Recognizing Actions in 3D Using Action-Snippets and Activated Simplices (AAAI 2016 · similarity 0.89)
- Unified Keypoint-Based Action Recognition Framework via Structured Keypoint Pooling (CVPR 2023 · similarity 0.86)
- Learning Discriminative Activated Simplices for Action Recognition (AAAI 2017 · similarity 0.86)
- A New Representation of Skeleton Sequences for 3D Action Recognition (CVPR 2017 · similarity 0.86)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.85)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 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.