Research map / Video / Video generation
SICAGE: Speaker-Independent Culture-Aware Gesture Generation using TED4C-L Dataset
Ariel Gjaci, Antonio Sgorbissa, Vittorio Murino
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 Fishr speaker-independent cultural embeddings (ablation) | ALaDiT · TED4C-L (speaker-disjoint test) | FGD: 1.60 -> 1.03 | Improved |
| add Fishr cultural embeddings (ablation) | ALaDiT · TED4C-L (speaker-disjoint test) | CE F1 (%): 43.41 -> 44.61 | Improved |
| add adversarial cultural embeddings instead of Fishr (ablation) | ALaDiT · TED4C-L (speaker-disjoint test) | FGD: 1.60 -> 1.53 | No clear effect |
| add adversarial cultural embeddings (ablation) | ALaDiT · TED4C-L (speaker-disjoint test) | Diversity: 109.50 -> 111.75 | Improved |
| use Fishr embeddings instead of one-hot cultural label (ablation) | ALaDiT · TED4C-L (speaker-disjoint test) | FGD: 1.63 -> 1.03 | Improved |
| add speaker-domain regularization (Fishr) vs NoDG (ablation) | ALaDiT · TED4C-L (speaker-disjoint test) | CE F1 (%): 43.18 -> 44.61 | Improved |
| add ALaDiT alignment losses (ablation) | ALaDiT/FI · TED4C-L (speaker-disjoint test) | FGD: 1.36 -> 1.03 | Improved |
| add ALaDiT alignment losses (ablation) | ALaDiT/FI · TED4C-L (speaker-disjoint test) | SRGR (%): 68.17 -> 68.09 | Hurt |
| add Fishr embeddings (ablation) | MDM · TED4C-L (speaker-disjoint test) | FGD: 15.58 -> 7.59 | Improved |
| add Fishr embeddings (ablation) | MDM · TED4C-L (speaker-disjoint test) | CE F1 (%): 38.57 -> 47.09 | Improved |
| add Fishr embeddings (ablation) | DSG+ · TED4C-L (speaker-disjoint test) | FGD: 2.76 -> 4.89 | Hurt |
| add explicit alignment to DSG+/FI (ablation) | DSG+ · TED4C-L (speaker-disjoint test) | FGD: 4.89 -> 2.52 | Improved |
Closest papers by meaning
- CoordSpeaker: Exploiting Gesture Captioning for Coordinated Caption-Empowered Co-Speech Gesture Generation (CVPR 2026 · similarity 0.89)
- ConvoFusion: Multi-Modal Conversational Diffusion for Co-Speech Gesture Synthesis (CVPR 2024 · similarity 0.88)
- Co$^{\mathbf{3}}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion (ICLR 2025 · similarity 0.86)
- EMAGE: Towards Unified Holistic Co-Speech Gesture Generation via Expressive Masked Audio Gesture Modeling (CVPR 2024 · similarity 0.86)
- Retrieving Semantics from the Deep: an RAG Solution for Gesture Synthesis (CVPR 2025 · similarity 0.86)
- The Language of Motion: Unifying Verbal and Non-verbal Language of 3D Human Motion (CVPR 2025 · similarity 0.86)
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.
