Research map / Image generation / Autoregressive and latent generation
MacDiff: Unified Skeleton Modeling with Masked Conditional Diffusion
Lehong Wu, Lilang Lin, Jiahang Zhang, Yiyang Ma, Jiaying Liu
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 |
|---|---|---|---|
| pretrain a masked conditional diffusion model instead of masked reconstruction (MAMP) | Transformer skeleton encoder, joint stream · NTU 60 xsub linear eval | accuracy: 84.9 -> 86.4 | Improved |
| MacDiff single joint stream vs three-stream ensemble contrastive methods | linear evaluation · NTU 60 xsub linear eval | accuracy: 84.3 -> 86.4 | Improved |
| MacDiff vs MAMP on NTU 120 xset | linear evaluation · NTU 120 xset linear eval | accuracy: 79.1 -> 80.2 | Improved |
| MacDiff vs MAMP on PKUMMD I | linear evaluation · PKUMMD I linear eval | accuracy: 92.2 -> 92.8 | Improved |
| MacDiff vs PCM3 on NTU 60 xview | linear evaluation · NTU 60 xview linear eval | accuracy: 90.4 -> 91.0 | Improved |
Closest papers by meaning
- Less is More: Compact-Token Masked Feature Learning for Skeleton Representation Learning (NeurIPS 2026 · similarity 0.90)
- MaskCLR: Attention-Guided Contrastive Learning for Robust Action Representation Learning (CVPR 2024 · similarity 0.89)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.88)
- Neuron: Learning Context-Aware Evolving Representations for Zero-Shot Skeleton Action Recognition (CVPR 2025 · similarity 0.87)
- Nonisotropic Gaussian Diffusion for Realistic 3D Human Motion Prediction (CVPR 2025 · similarity 0.87)
- Universal Skeleton Understanding via Differentiable Rendering and MLLMs (ICML 2026 · 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.
