PulseAugur
EN
LIVE 10:30:06

H-Flow model estimates human motion from video using physics priors

By PulseAugur Editorial · [2 sources] ·

Researchers have developed H-Flow, a novel self-supervised method for estimating human scene flow from monocular video. This approach integrates skeletal kinematics with surface deformation, overcoming limitations of existing models. H-Flow leverages physics-inspired priors and a new synthetic benchmark, DynAct4D, to achieve state-of-the-art performance and generalize to real-world videos. AI

IMPACT Introduces a new method for dense human motion estimation, potentially improving animation and virtual reality applications.

RANK_REASON The cluster contains an academic paper detailing a new model and benchmark.

Read on arXiv cs.CV →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

H-Flow model estimates human motion from video using physics priors

COVERAGE [2]

  1. arXiv cs.CV TIER_1 English(EN) · Zhanbo Huang, Xiaoming Liu, Yu Kong ·

    H-Flow: Self-supervised Human Scene Flow via Physics-inspired Joint Multi-modal Learning

    arXiv:2605.22629v1 Announce Type: new Abstract: Parametric human models capture global pose but cannot represent the non-rigid surface dynamics of clothing and soft tissue. Generic scene flow estimates dense motion but breaks down on articulated bodies, where pixel-level supervis…

  2. arXiv cs.CV TIER_1 English(EN) · Yu Kong ·

    H-Flow: Self-supervised Human Scene Flow via Physics-inspired Joint Multi-modal Learning

    Parametric human models capture global pose but cannot represent the non-rigid surface dynamics of clothing and soft tissue. Generic scene flow estimates dense motion but breaks down on articulated bodies, where pixel-level supervision is also intractable to acquire. We introduce…

RELATED ENTITIES

RELATED TOPICS