Research map / Video / Video generation
DIM: Dyadic Interaction Modeling for Social Behavior Generation
Minh Tran, Di Chang, Maksim Siniukov, Mohammad Soleymani
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 dyadic interaction pre-training with joint speaker-listener modeling (ablation) | DIM-Listener · ViCo | FD (Exp): 33.84 -> 24.15 | Improved |
| pre-train L2L on CANDOR as well | L2L · ViCo | FD (Exp): 33.93 -> 31.03 | Improved |
| use VQ-VAE discrete codes instead of continuous prediction (ablation) | DIM-Listener · ViCo | FD (Exp): 42.52 -> 24.15 | Improved |
| unfreeze listener VQ decoder and integrate into motion prediction (ablation) | DIM-Listener · ViCo | FD (Exp): 39.63 -> 24.15 | Improved |
| add contrastive loss between speaker and listener features (ablation) | DIM-Listener · ViCo | FD (Exp): 28.47 -> 24.15 | Improved |
| add two-branch speaker-listener joint modeling (S-L) (ablation) | DIM-Listener without contrastive loss · ViCo | FD (Exp): 30.92 -> 28.47 | Improved |
| add contrastive loss between speaker and listener features (ablation) | DIM-Listener · LM_Listener | FD (Exp): 17.92 -> 12.94 | Improved |
| use both audio and visual speaker inputs instead of single modality (ablation) | DIM-Listener · ViCo | FD (Exp): 32.8 -> 23.9 | Improved |
| use causal masks for online generation (ablation) | DIM-Listener · ViCo | FD (Exp): 23.9 -> 28.4 | Hurt |
| apply DIM pre-trained model to speaker motion generation (DIM-Speaker) | DIM-Speaker · BiWi | LVE (x10^-4): 5.40 -> 3.86 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FaceFormer — FaceFormer: Speech-Driven 3D Facial Animation With Transformers (CVPR 2022) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- INFP: Audio-Driven Interactive Head Generation in Dyadic Conversations (CVPR 2025 · similarity 0.90)
- Full-Duplex Speech-Motion Model for Dyadic Interaction (NeurIPS 2026 · similarity 0.87)
- The Language of Motion: Unifying Verbal and Non-verbal Language of 3D Human Motion (CVPR 2025 · similarity 0.87)
- GLARE: Generating Listening Heads with Appropriate REactions (NeurIPS 2026 · similarity 0.86)
- DEEPTalk: Dynamic Emotion Embedding for Probabilistic Speech-Driven 3D Face Animation (AAAI 2025 · similarity 0.86)
- Motion-Agent: A Conversational Framework for Human Motion Generation with LLMs (ICLR 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.
