Research map / 3D vision / 3D generation
HO-Flow: Generalizable Hand-Object Interaction Generation with Latent Flow Matching
Zerui Chen, Rolandos Alexandros Potamias, Shizhe Chen, Jiankang Deng, Cordelia Schmid, Stefanos Zafeiriou
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 object kinematics: world-frame object point clouds with 6D object poses (ablation) | Inter-VAE · GRAB | E_o: 16.34 -> 8.93 | Improved |
| add hand-joint-frame transformation of object points (ablation) | Inter-VAE · GRAB | E_v^l: 20.51 -> 16.59 | Improved |
| use separate latents for hand and object (ablation) | Inter-VAE · GRAB | E_v^r: 21.69 -> 17.29 | Improved |
| pre-train Inter-VAE on GraspXL synthetic trajectories (ablation) | Inter-VAE · GRAB | E_v^r: 17.29 -> 9.13 | Improved |
| replace frame-independent latent with Inter-VAE spatial-temporal latent (ablation) | HO-Flow generator · GRAB | Phy: 88.76 -> 92.69 | Improved |
| add auto-regressive context-aware transformer (ablation) | HO-Flow generator · GRAB | Phy: 92.69 -> 97.94 | Improved |
| add latent representation over raw pose-space generation (ablation) | HO-Flow generator · GRAB | Phy: 84.89 -> 88.76 | Improved |
| add GraspXL pre-training to generator (ablation) | HO-Flow · GRAB | Phy: 97.94 -> 98.25 | Improved |
| add GraspXL pre-training to generator | HO-Flow · OakInk | Phy: 83.62 -> 89.76 | Improved |
| add GraspXL pre-training to generator | HO-Flow · DexYCB | Phy: 90.77 -> 95.41 | Improved |
| use flow matching instead of DDPM (ablation) | HO-Flow · GRAB | Phy: 96.00 -> 98.25 | Improved |
| increase inference steps from 6 to 18 (ablation) | HO-Flow · GRAB | Phy: 90.63 -> 98.25 | Improved |
Closest papers by meaning
- FlowHOI: Flow-based Semantics-Grounded Generation of Hand-Object Interactions for Dexterous Robot Manipulation (NeurIPS 2026 · similarity 0.92)
- HOIGPT: Learning Long-Sequence Hand-Object Interaction with Language Models (CVPR 2025 · similarity 0.91)
- SViMo: Synchronized Diffusion for Video and Motion Generation in Hand-object Interaction Scenarios (NeurIPS 2025 · similarity 0.90)
- How Do I Do That? Synthesizing 3D Hand Motion and Contacts for Everyday Interactions (CVPR 2025 · similarity 0.90)
- CG-HOI: Contact-Guided 3D Human-Object Interaction Generation (CVPR 2024 · similarity 0.90)
- LatentHOI: On the Generalizable Hand Object Motion Generation with Latent Hand Diffusion. (CVPR 2025 · similarity 0.90)
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.
