Research map / 3D vision / 3D generation
PolySLGen: Online Multimodal Speaking-Listening Reaction Generation in Polyadic Interaction
Zhi-Yi Lin, Thomas Markhorst, Jouh Yeong Chew, Xucong Zhang
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 pose fusion module (no social cue) (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 153.3 -> 148.2 | Improved |
| add pose fusion module (no social cue) (ablation) | PolySLGen · DnD Group Gesture | State AP: 0.60 -> 0.66 | Improved |
| add social cue encoder alone (no pose fusion) (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 153.3 -> 152.5; WER 1.454 -> 1.715 | Mixed |
| add social cue encoder on top of pose fusion (ablation) | PolySLGen · DnD Group Gesture | Root (mm): 116.7 -> 108.7 | Improved |
| add social cue encoder on top of pose fusion (ablation) | PolySLGen · DnD Group Gesture | State AP: 0.66 -> 0.67 | Improved |
| add verbal (speaking participants') motion observation (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 202.5 -> 175.5 | Improved |
| add non-verbal (non-speaking participants') motion observation (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 175.5 -> 153.3 | Improved |
| remove up to 1 non-target participant at inference (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 144.9 -> 176.3 | Hurt |
| generate speech before motion instead of motion before speech (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 154.9 -> 144.9 | Improved |
| use social cue embedding length 2 instead of 1 (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 152.3 -> 144.9 | Improved |
| use social cue embedding length 2 instead of 4 (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 149.8 -> 144.9 | Improved |
| use translation plus 6D rotations instead of 3D keypoints (ablation) | PolySLGen · DnD Group Gesture | MPJPE (mm): 150.8 -> 144.9 | Improved |
Closest papers by meaning
- INFP: Audio-Driven Interactive Head Generation in Dyadic Conversations (CVPR 2025 · similarity 0.89)
- OmniResponse: Online Multimodal Conversational Response Generation in Dyadic Interactions (NeurIPS 2025 · similarity 0.87)
- ConvoFusion: Multi-Modal Conversational Diffusion for Co-Speech Gesture Synthesis (CVPR 2024 · similarity 0.87)
- Full-Duplex Speech-Motion Model for Dyadic Interaction (NeurIPS 2026 · similarity 0.87)
- DualTalk: Dual-Speaker Interaction for 3D Talking Head Conversations (CVPR 2025 · similarity 0.87)
- Let Them Talk: Audio-Driven Multi-Person Conversational Video Generation (NeurIPS 2025 · 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.
