Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
ProjFlow: Projection Sampling with Flow Matching for Zero-Shot Exact Spatial Motion Control
Akihisa Watanabe, Qing Yu, Edgar Simo-Serra, Kent Fujiwara
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 Euclidean metric with kinematics-aware metric (ablation) | ProjFlow · HumanML3D inpainting | FID: 1.152 -> 0.097 | Improved |
| add stochastic recomposition (noise mixing) (ablation) | ProjFlow · HumanML3D inpainting | FID: 3.429 -> 0.097 | Improved |
| add time-varying pseudo-observations (ablation) | ProjFlow · HumanML3D inpainting | FID: 0.880 -> 0.097 | Improved |
| replace Euclidean metric with kinematics-aware metric (ablation) | ProjFlow · HumanML3D inpainting | control error: 0.0000 for all variants | No clear effect |
| use ProjFlow projection instead of DNO noise optimization | ACMDM-S-PS22 · pelvis control | FID: 0.151 -> 0.107 | Improved |
| use ProjFlow instead of trained ControlNet branch | ACMDM-S-PS22 · pelvis control | FID: 0.067 -> 0.107 | Hurt |
| use ProjFlow instead of CtrlNet | ACMDM-S-PS22 · pelvis control | average error: 0.0100 -> 0.0000 | Improved |
| use ProjFlow instead of Sketch2Anim | 2D-to-3D lifting · HumanML3D Average | FID: 0.525 -> 0.349 | Improved |
| use ProjFlow instead of Sketch2Anim | 2D-to-3D lifting · HumanML3D Average | Foot Skating: 0.103 -> 0.146 | Hurt |
| use ProjFlow instead of Sketch2Anim | 2D-to-3D lifting · HumanML3D Average | Avg. Err.-3D: 0.134 -> 0.331 | Hurt |
Closest papers by meaning
- HardFlow: Hard-Constrained Sampling for Flow-Matching Models via Trajectory Optimization (TPAMI 2026 · similarity 0.87)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 2025 · similarity 0.86)
- FlowPolicy: Enabling Fast and Robust 3D Flow-Based Policy via Consistency Flow Matching for Robot Manipulation (AAAI 2025 · similarity 0.86)
- Boximator: Generating Rich and Controllable Motions for Video Synthesis (ICML 2024 · similarity 0.85)
- CoDA: Coordinated Diffusion Noise Optimization for Whole-Body Manipulation of Articulated Objects (NeurIPS 2025 · similarity 0.85)
- MotionPRO: Exploring the Role of Pressure in Human MoCap and Beyond (CVPR 2025 · similarity 0.85)
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.
