Research map / 3D vision / 3D generation
Object Pop-Up: Can We Infer 3D Objects and Their Poses From Human Interactions Alone?
Ilya A. Petrov, Riccardo Marin, Julian Chibane, Gerard Pons-Moll
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 |
|---|---|---|---|
| replace nearest-neighbor retrieval with human-point-cloud object-pose prediction | Object Pop-Up · GRAB | object center error (E_c): 0.0362 -> 0.0237 | Improved |
| replace nearest-neighbor retrieval with human-point-cloud object-pose prediction | Object Pop-Up · GRAB | vertex-to-vertex error (E_v2v): 0.1445 -> 0.0943 | Improved |
| replace nearest-neighbor retrieval with human-point-cloud object-pose prediction | Object Pop-Up · BEHAVE | object center error (E_c): 0.0802 -> 0.0663 | Improved |
| replace nearest-neighbor retrieval with human-point-cloud object-pose prediction | Object Pop-Up · BEHAVE | vertex-to-vertex error (E_v2v): 0.3445 -> 0.2900 | Improved |
| add temporal smoothing to object-class predictions (ablation) | Object Pop-Up with class prediction · GRAB | accuracy: 20.80 -> 30.15 | Improved |
| add temporal smoothing to object-class predictions (ablation) | Object Pop-Up with class prediction · BEHAVE | accuracy: 24.16 -> 49.28 | Improved |
| use hand points only instead of full-body input (ablation) | Object Pop-Up · GRAB | vertex-to-vertex error (E_v2v): 0.0943 -> 0.0871 | Improved |
| use SMPL input with hands in rest pose instead of articulated hands (ablation) | Object Pop-Up · GRAB | vertex-to-vertex error (E_v2v): 0.0943 -> 0.1009 | Hurt |
| use SMPL input with hands in rest pose instead of articulated hands (ablation) | Object Pop-Up · BEHAVE | vertex-to-vertex error (E_v2v): 0.2900 -> 0.2877 | No clear effect |
| predict global rotation and translation directly instead of point-wise offsets (ablation) | Object Pop-Up · GRAB | vertex-to-vertex error (E_v2v): 0.0943 -> 0.2556 | Hurt |
| remove local-neighborhood features (ablation) | Object Pop-Up · GRAB | vertex-to-vertex error (E_v2v): 0.0943 -> 0.1112 | Hurt |
| remove local-neighborhood features (ablation) | Object Pop-Up · BEHAVE | vertex-to-vertex error (E_v2v): 0.2900 -> 0.2868 | No clear effect |
Closest papers by meaning
- InteractAnything: Zero-shot Human Object Interaction Synthesis via LLM Feedback and Object Affordance Parsing (CVPR 2025 · similarity 0.88)
- GREAT: Geometry-Intention Collaborative Inference for Open-Vocabulary 3D Object Affordance Grounding (CVPR 2025 · similarity 0.88)
- Evaluating Multiview Object Consistency in Humans and Image Models (NeurIPS 2024 · similarity 0.87)
- Vision-Guided Action: Enhancing 3D Human Motion Prediction with Gaze-informed Affordance in 3D Scenes (CVPR 2025 · similarity 0.87)
- PICO: Reconstructing 3D People In Contact with Objects (CVPR 2025 · similarity 0.86)
- EgoChoir: Capturing 3D Human-Object Interaction Regions from Egocentric Views (NeurIPS 2024 · 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.
