AI Paper Atlas

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

CVPR 2026 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Transductive and semantic-dictionary zero-shot · 3 citations (OpenAlex)

Read the paper

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 changedWhereMeasuredEffect
replace Euclidean metric with kinematics-aware metric (ablation)ProjFlow · HumanML3D inpaintingFID: 1.152 -> 0.097Improved
add stochastic recomposition (noise mixing) (ablation)ProjFlow · HumanML3D inpaintingFID: 3.429 -> 0.097Improved
add time-varying pseudo-observations (ablation)ProjFlow · HumanML3D inpaintingFID: 0.880 -> 0.097Improved
replace Euclidean metric with kinematics-aware metric (ablation)ProjFlow · HumanML3D inpaintingcontrol error: 0.0000 for all variantsNo clear effect
use ProjFlow projection instead of DNO noise optimizationACMDM-S-PS22 · pelvis controlFID: 0.151 -> 0.107Improved
use ProjFlow instead of trained ControlNet branchACMDM-S-PS22 · pelvis controlFID: 0.067 -> 0.107Hurt
use ProjFlow instead of CtrlNetACMDM-S-PS22 · pelvis controlaverage error: 0.0100 -> 0.0000Improved
use ProjFlow instead of Sketch2Anim2D-to-3D lifting · HumanML3D AverageFID: 0.525 -> 0.349Improved
use ProjFlow instead of Sketch2Anim2D-to-3D lifting · HumanML3D AverageFoot Skating: 0.103 -> 0.146Hurt
use ProjFlow instead of Sketch2Anim2D-to-3D lifting · HumanML3D AverageAvg. Err.-3D: 0.134 -> 0.331Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy