Research map / Video / Video segmentation and tracking
Spatial-Temporal Graph Mamba for Music-Guided Dance Video Synthesis
Hao Tang, Ling Shao, Zhenyu Zhang, Luc Van Gool, Nicu Sebe
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 SG-SSM (spatial graph SSM) (ablation) | MDVS-style baseline · Music-to-skeleton (k-pop/ballet/popping) | PFD: 653.8 -> 428.3 | Improved |
| add TGF-SSM (temporal forward SSM) (ablation) | MDVS-style baseline · Music-to-skeleton (k-pop/ballet/popping) | PFD: 653.8 -> 326.9 | Improved |
| add TGB-SSM (temporal backward SSM) (ablation) | MDVS-style baseline · Music-to-skeleton (k-pop/ballet/popping) | PFD: 653.8 -> 318.6 | Improved |
| add TGB-SSM to TGF-SSM (bidirectional temporal modeling) (ablation) | MDVS-style baseline · Music-to-skeleton (k-pop/ballet/popping) | PFD: 326.9 -> 130.7 | Improved |
| add SG-SSM to bidirectional temporal SSMs (ablation) | STG-Mamba · Music-to-skeleton (k-pop/ballet/popping) | PFD: 130.7 -> 86.2 | Improved |
| add SG-SSM to bidirectional temporal SSMs (ablation) | STG-Mamba · Music-to-skeleton (k-pop/ballet/popping) | BC: 0.629 -> 0.684 | Improved |
| add SG-SSM to bidirectional temporal SSMs (ablation) | STG-Mamba · Music-to-skeleton (k-pop/ballet/popping) | PVar: 0.761 -> 0.875 | Improved |
| add forward self-supervised regularization (FSR) (ablation) | Vid2Vid baseline · Skeleton-to-video (collected dataset) | FVD: 81.5 -> 57.3 | Improved |
| add backward self-supervised regularization (BSR) (ablation) | Vid2Vid baseline · Skeleton-to-video (collected dataset) | FVD: 81.5 -> 55.2 | Improved |
| add BSR to FSR (ablation) | Vid2Vid baseline · Skeleton-to-video (collected dataset) | FID: 42.89 -> 35.17 | Improved |
| add BSR to FSR (ablation) | Vid2Vid baseline · Skeleton-to-video (collected dataset) | LPIPS: 0.1785 -> 0.1529 | Improved |
| use Mamba-based STGM blocks instead of Transformer (EDGE) | STG-Mamba · Music-to-skeleton (k-pop/ballet/popping) | Inference time (s): 4.8 -> 2.2 | Improved |
Closest papers by meaning
- BRACE: The Breakdancing Competition Dataset for Dance Motion Synthesis (ECCV 2022 · similarity 0.86)
- DanceAnyWay: Synthesizing Beat-Guided 3D Dances with Randomized Temporal Contrastive Learning (AAAI 2024 · similarity 0.86)
- Convolutional Sequence Generation for Skeleton-Based Action Synthesis (ICCV 2019 · similarity 0.86)
- DanceEditor: Towards Iterative Editable Music-driven Dance Generation with Open-Vocabulary Descriptions (ICCV 2025 · similarity 0.86)
- X-Dancer: Expressive Music to Human Dance Video Generation (ICCV 2025 · similarity 0.86)
- DanceCamera3D: 3D Camera Movement Synthesis with Music and Dance (CVPR 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.
