Research map / 3D vision / Human pose estimation
Synthesizing Long-Term 3D Human Motion and Interaction in 3D Scenes
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use a hierarchical route-and-pose motion model instead of Route+CVAE | Long-term motion synthesis · PROX unseen scenes | MPJPE: 294.2 -> 191.6 | Helped |
| Use a hierarchical route-and-pose motion model instead of Route+CVAE | Long-term motion synthesis · PROX | human naturalness score, 2 seconds: 2.23$\pm$1.10 -> 3.39$\pm$1.11 | Helped |
| Apply geometric optimization with foot, collision, contact, and smoothness constraints (ablation) | Ours motion synthesis · PROX | non-collision: 97.53 -> 99.91 | Helped |
| Apply geometric optimization with foot, collision, contact, and smoothness constraints (ablation) | Ours motion synthesis · PROX unseen scenes | MPJPE: 191.6 -> 219.1 | Hurt |
| Add scene point-cloud information to the motion networks (ablation) | RouteNet(t)+PoseNet · PROX unseen scenes | transl: 7.54 -> 7.31 | Helped |
| Add scene point-cloud information to the motion networks (ablation) | RouteNet(t+r)+PoseNet · PROX unseen scenes | MPJPE: 201.6 -> 195.8 | Helped |
| Predict translation and orientation in RouteNet rather than translation alone (ablation) | RouteNet+PoseNet · PROX unseen scenes | orientation: 11.65 -> 9.71 | Helped |
| Provide the predicted route as input to PoseNet (ablation) | Motion synthesis network · PROX unseen scenes | MPJPE: 235.8 -> 195.8 | Helped |
| Apply geometric optimization to Route+CVAE (ablation) | Route+CVAE · PROX | non-collision: 98.57 -> 99.88 | Helped |
| Apply geometric optimization to Route+CVAE (ablation) | Route+CVAE · PROX | contact: 88.68 -> 99.17 | Hurt |
| Apply geometric optimization to HSNMS (ablation) | HSNMS · MP3D | non-collision: 95.29 -> 99.57 | Helped |
| Use a one-network end-to-end route-and-pose predictor instead of separate route and pose networks (ablation) | BaseNet · PROX unseen scenes | MPJPE: 195.8 -> 226.2 | Hurt |
Closest papers by meaning
- Move as You Say Interact as You Can: Language-guided Human Motion Generation with Scene Affordance (CVPR 2024 · similarity 0.91)
- Vision-Guided Action: Enhancing 3D Human Motion Prediction with Gaze-informed Affordance in 3D Scenes (CVPR 2025 · similarity 0.90)
- AffordBot: 3D Fine-grained Embodied Reasoning via Multimodal Large Language Models (NeurIPS 2025 · similarity 0.89)
- NIFTY: Neural Object Interaction Fields for Guided Human Motion Synthesis (CVPR 2024 · similarity 0.88)
- SeqAfford: Sequential 3D Affordance Reasoning via Multimodal Large Language Model (CVPR 2025 · similarity 0.88)
- GREAT: Geometry-Intention Collaborative Inference for Open-Vocabulary 3D Object Affordance Grounding (CVPR 2025 · similarity 0.88)
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.