Research map / Label-efficient and robust learning / Recognition and long-tail
Generative Multi-View Human Action Recognition
Lichen Wang, Zhengming Ding, Zhiqiang Tao, Yunyu Liu, Yun Fu
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 GAN instead of a mapping module (ablation) | GMVAR · UWA3D (R+D) | accuracy (%): 74.52 -> 76.28 | Improved |
| use GAN instead of a mapping module (ablation) | GMVAR · DHA (R+D) | accuracy (%): 88.07 -> 88.72 | Improved |
| use GAN instead of a mapping module (ablation) | GMVAR · MHAD (R+D) | accuracy (%): 98.23 -> 98.94 | Improved |
| use VCDN instead of averaging view labels (ablation) | GMVAR (no generator) · UWA3D (R+D) | accuracy (%): 71.84 -> 74.07 | Improved |
| use VCDN instead of concatenating view labels (ablation) | GMVAR (no generator) · UWA3D (R+D) | accuracy (%): 70.38 -> 74.07 | Improved |
| use VCDN instead of concatenating original features (ablation) | GMVAR (no generator) · UWA3D (R+D) | accuracy (%): 69.22 -> 74.07 | Improved |
| use VCDN instead of averaging view labels (ablation) | GMVAR (no generator) · DHA (R+D) | accuracy (%): 83.28 -> 84.32 | Improved |
| use VCDN instead of 2-layer MLP fusion (ablation) | GMVAR · UWA3D (R+D) | accuracy (%): 74.70 -> 76.28 | Improved |
| use VCDN instead of 4-layer MLP fusion (ablation) | GMVAR · MHAD (R+D) | accuracy (%): 95.76 -> 98.94 | Improved |
| fuse encoder features by concatenation instead of using RGB view only (ablation) | GMVAR (no generator) · UWA3D | accuracy (%): 69.18 -> 68.78 | Hurt |
| learned weighted label fusion instead of averaging (ablation) | GMVAR (no generator) · UWA3D (R+D) | accuracy (%): 71.84 -> 71.15 | No clear effect |
| add generated view to complete missing depth view (D->R) | GMVAR · UWA, MHAD, DHA | accuracy (%): +3% average | Improved |
Closest papers by meaning
- DVANet: Disentangling View and Action Features for Multi-View Action Recognition (AAAI 2024 · similarity 0.88)
- Multi-view Action Recognition using Cross-view Video Prediction (ECCV 2020 · similarity 0.87)
- Dividing and Aggregating Network for Multi-view Action Recognition (ECCV 2018 · similarity 0.87)
- Generative View-Correlation Adaptation for Semi-Supervised Multi-View Learning (ECCV 2020 · similarity 0.87)
- PM-GANs: Discriminative Representation Learning for Action Recognition Using Partial-modalities (ECCV 2018 · similarity 0.86)
- Semi-Supervised Multi-View Multi-Label Learning with View-Specific Transformer and Enhanced Pseudo-Label (AAAI 2025 · 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.
