Research map / Video / Video generation
DyaDiT: A Multi-Modal Diffusion Transformer for Socially Favorable Dyadic Gesture Generation
Yichen Peng, Jyun-Ting Song, Siyeol Jung, Ulsan National Institute of Science & Technology blank, Ruofan Liu, Haiyang Liu, Xuangeng Chu, Ruicong Liu, Erwin Wu, Hideki Koike, Kris Kitani
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 ORCA audio fusion instead of concatenation (ablation) | DyaDiT · Seamless Interaction | FD Static: 7.32 -> 6.40 | Improved |
| add ORCA audio fusion instead of concatenation (ablation) | DyaDiT · Seamless Interaction | Diversity Static: 23.57 -> 27.46 | Mixed |
| use ORCA instead of standard cross-attention (ablation) | DyaDiT · Seamless Interaction | FD Static: 7.82 -> 6.40 | Improved |
| use ORCA instead of standard cross-attention (ablation) | DyaDiT · Seamless Interaction | Diversity Static: 18.87 -> 27.46 | Improved |
| add discrete motion dictionary (ablation) | DyaDiT · Seamless Interaction | Diversity Static: 18.34 -> 27.46 | Improved |
| use discrete instead of continuous motion dictionary (ablation) | DyaDiT · Seamless Interaction | Diversity Static: 21.47 -> 27.46 | Improved |
| add self motion conditioning (ablation) | DyaDiT · Seamless Interaction | FD Kinetic: 1.48 -> 1.37 | Improved |
| add self motion conditioning (ablation) | DyaDiT · Seamless Interaction | Diversity Static: 26.43 -> 27.46 | No clear effect |
| add relationship and personality conditioning (ablation) | DyaDiT · Seamless Interaction | FD Static: 7.40 -> 6.40 | Improved |
| use true instead of random relationship/personality labels (ablation) | DyaDiT · Seamless Interaction | FD Static: 8.24 -> 6.40 | Improved |
| use true instead of random relationship/personality labels (ablation) | DyaDiT · Seamless Interaction | Diversity Kinetic: 1.43 -> 1.38 | Hurt |
| use DyaDiT instead of ConvoFusion | dyadic gesture generation · Seamless Interaction | FD Static: 9.22 -> 6.40 | Improved |
Closest papers by meaning
- ConvoFusion: Multi-Modal Conversational Diffusion for Co-Speech Gesture Synthesis (CVPR 2024 · similarity 0.90)
- INFP: Audio-Driven Interactive Head Generation in Dyadic Conversations (CVPR 2025 · similarity 0.89)
- DiffSHEG: A Diffusion-Based Approach for Real-Time Speech-driven Holistic 3D Expression and Gesture Generation (CVPR 2024 · similarity 0.89)
- Full-Duplex Speech-Motion Model for Dyadic Interaction (NeurIPS 2026 · similarity 0.89)
- Co$^{\mathbf{3}}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion (ICLR 2025 · similarity 0.88)
- Co-Speech Gesture Video Generation via Motion-Decoupled Diffusion Model (CVPR 2024 · similarity 0.88)
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.
