Research map / Video / Video generation
GestureHYDRA: Semantic Co-speech Gesture Synthesis via Hybrid Modality Diffusion Transformer and Cascaded-Synchronized Retrieval-Augmented Generation
Quanwei Yang, Luying Huang, Kaisiyuan Wang, Jiazhi Guan, Shengyi He, Fengguo Li, Hang Zhou, Lingyun Yu, Yingying Li, Haocheng Feng, Hongtao Xie
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 motion style conditioning (ablation) | GestureHYDRA · Streamer (unseen identity) | FGD: 20.31 -> 15.43 | Improved |
| add motion style conditioning (ablation) | GestureHYDRA · Streamer (unseen identity) | SMD-DTW: 29.51 -> 27.73 | Improved |
| add the mask strategy (ablation) | GestureHYDRA · Streamer (unseen identity) | FGD: 15.76 -> 15.43 | Improved |
| add the mask strategy (ablation) | GestureHYDRA · Streamer (unseen identity) | SMD-DTW: 29.71 -> 27.73 | Improved |
| add 3D keypoint loss (ablation) | GestureHYDRA · Streamer (unseen identity) | FGD: 14.80 -> 15.43 | Hurt |
| add 3D keypoint loss (ablation) | GestureHYDRA · Streamer (unseen identity) | SMD-L1: 0.154 -> 0.143 | Improved |
| add key gesture injection at ASR midpoint (vanilla) vs none (ablation) | GestureHYDRA · Streamer (failed-activation samples) | SMD-DTW: 30.46 -> 27.45 | Improved |
| use adaptive injection instead of vanilla injection (ablation) | GestureHYDRA · Streamer (failed-activation samples) | SMD-DTW: 27.45 -> 26.88 | Improved |
| use adaptive injection instead of vanilla injection (ablation) | GestureHYDRA · Streamer (failed-activation samples) | SMD-L1: 0.155 -> 0.138 | Improved |
| use GestureHYDRA instead of DSG | co-speech gesture synthesis · Streamer (seen identity) | FGD: 54.59 -> 3.24 | Improved |
| use GestureHYDRA instead of DSG | co-speech gesture synthesis · Streamer (unseen identity) | SAR: 68.77% -> 81.36% | Improved |
| use GestureHYDRA instead of Probtalk | co-speech gesture synthesis · SHOW | FGD: 5.46 -> 3.68 | Improved |
Closest papers by meaning
- Co$^{\mathbf{3}}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion (ICLR 2025 · similarity 0.91)
- ConvoFusion: Multi-Modal Conversational Diffusion for Co-Speech Gesture Synthesis (CVPR 2024 · similarity 0.91)
- Retrieving Semantics from the Deep: an RAG Solution for Gesture Synthesis (CVPR 2025 · similarity 0.91)
- DiffSHEG: A Diffusion-Based Approach for Real-Time Speech-driven Holistic 3D Expression and Gesture Generation (CVPR 2024 · similarity 0.90)
- Co-Speech Gesture Video Generation via Motion-Decoupled Diffusion Model (CVPR 2024 · similarity 0.89)
- Chain of Generation: Multi-Modal Gesture Synthesis via Cascaded Conditional Control (AAAI 2024 · similarity 0.89)
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.
