Research map / 3D vision / 3D generation
3D Human Action Representation Learning via Cross-View Consistency Pursuit
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 |
|---|---|---|---|
| Add cross-view consistency learning across three skeleton views | SkeletonCLR · NTU-60 xsub | accuracy: 75.0 -> 77.8 | Helped |
| Add cross-view consistency learning across three skeleton views | SkeletonCLR · NTU-60 xview | accuracy: 79.8 -> 83.4 | Helped |
| Add top-1 high-confidence sample mining (ablation) | 2s-CrosSCLR · NTU-60 xsub | accuracy: 70.5 -> 74.5 | Helped |
| Increase top-K from 1 to 3 (ablation) | 2s-CrosSCLR · NTU-60 xsub | accuracy: 74.5 -> 73.7 | Hurt |
| Increase top-K from 1 to 10 (ablation) | 2s-CrosSCLR · NTU-60 xview | accuracy: 82.1 -> 69.9 | Hurt |
| Add single-view knowledge mining (ablation) | SkeletonCLR · NTU-60 xsub | accuracy: 68.3 -> 69.3 | Helped |
| Add embedding context to cross-view learning (ablation) | CrosSCLR without embedding context · NTU-60 xsub | accuracy: 71.4 -> 72.9 | Helped |
| Add embedding context to cross-view learning (ablation) | CrosSCLR without embedding context · NTU-60 xview | accuracy: 78.5 -> 79.9 | Helped |
| Add cross-view consistent knowledge mining | SkeletonCLR, joint view · NTU-60 xview | accuracy: 76.4 -> 79.9 | Helped |
| Add temporal crop augmentation with padding ratio 6 (ablation) | SkeletonCLR, joint view, shear amplitude 0.5 · NTU-60 xview | accuracy: 68.8 -> 76.4 | Helped |
| Increase shear amplitude from 0.5 to 1.0 with no crop (ablation) | SkeletonCLR, joint view · NTU-60 xsub | accuracy: 66.3 -> 62.0 | Hurt |
| Use the pretrained CrosSCLR encoder and finetune it | 3s-ST-GCN* · NTU-120 xsub | accuracy: 77.2 -> 80.5 | Helped |
Closest papers by meaning
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 2024 · similarity 0.85)
- MaskCLR: Attention-Guided Contrastive Learning for Robust Action Representation Learning (CVPR 2024 · similarity 0.84)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.84)
- Navigating Open Set Scenarios for Skeleton-Based Action Recognition (AAAI 2024 · similarity 0.84)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.83)
- Cross-view and Cross-pose Completion for 3D Human Understanding (CVPR 2024 · similarity 0.83)
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.