Research map / 3D vision / Human pose estimation
SpatialHand: Generative Object Manipulation from 3D Prespective
Zehan Wang, Jialei Wang, Siyu Chen, Ziang Zhang, Luping Liu, Xize Cheng, Kaihang Pan, Hengshuang Zhao, Zhou Zhao
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 textual caption to visual reference (ablation) | SpatialHand · 3D-aware insertion | DINO: 78.8 -> 81.7 | Improved |
| add textual caption to visual reference (ablation) | SpatialHand · 3D-aware insertion | Acc@30: 46.8 -> 52.0 | Improved |
| add textual caption to visual reference (ablation) | SpatialHand · 3D-aware insertion | AbsRel: 18.5 -> 19.8 | Hurt |
| add geometry-aware composition (ablation) | SpatialHand · 3D-aware insertion | AbsRel: 25.5 -> 19.8 | Improved |
| add geometry-aware composition (ablation) | SpatialHand · 3D-aware insertion | DINO: 80.2 -> 81.7 | Improved |
| add geometry-aware composition (ablation) | SpatialHand · 3D-aware insertion | Acc@30: 48.8 -> 52.0 | Improved |
| add stage 1 novel view synthesis fine-tuning (ablation) | SpatialHand · 3D-aware insertion | Acc@30: 28.7 -> 52.0 | Improved |
| add stage 1 novel view synthesis fine-tuning (ablation) | SpatialHand · 3D-aware insertion | DINO: 79.8 -> 81.7 | Improved |
| add stage 1 novel view synthesis fine-tuning (ablation) | SpatialHand · 3D-aware insertion | AbsRel: 18.2 -> 19.8 | No clear effect |
| use SpatialHand pose conditions instead of GPT-4o text instructions | SpatialHand vs GPT-4o · 3D-aware insertion (visual+textual) | Acc@30: 20.2 -> 52.0 | Improved |
| use SpatialHand pose conditions instead of GPT-4o text instructions | SpatialHand vs GPT-4o · 3D-aware insertion (visual+textual) | AbsRel: 38.6 -> 19.8 | Improved |
| use SpatialHand instead of GPT-4o for occlusion movement | SpatialHand vs GPT-4o · 3D-aware movement (occlusion) | VLM-Acc: 59.5 -> 82.6 | Improved |
Closest papers by meaning
- Interactive3D: Create What You Want by Interactive 3D Generation (CVPR 2024 · similarity 0.88)
- Localizing, Structuring, and Rendering: Bridging 3D and 2D Vision-Language-Action Models for Robotic Manipulation (CVPR 2026 · similarity 0.87)
- Image Sculpting: Precise Object Editing with 3D Geometry Control (CVPR 2024 · similarity 0.87)
- SceneDesigner: Controllable Multi-Object Image Generation with 9-DoF Pose Manipulation (NeurIPS 2025 · similarity 0.87)
- ObjectMover: Generative Object Movement with Video Prior (CVPR 2025 · similarity 0.86)
- VideoHandles: Editing 3D Object Compositions in Videos Using Video Generative Priors (CVPR 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.
