Cloth-Splatting: RGB画像からの3D布状態推定
Cloth-Splatting: 3D Cloth State Estimation from RGB Supervision
RGB画像のみを教師信号として、3Dガウシアンスプラッティングとメッシュ表現を組み合わせ、布の3D状態を予測・更新する手法を提案。推定精度の向上と収束時間の短縮を実現した。
著者: Alberta Longhini, Marcel Büsching, Bardienus P. Duisterhof, Jens Lundell, Jeffrey Ichnowski, Mårten Björkman, Danica Kragic
分類: cs.CV, cs.RO
原文アブストラクト
We introduce Cloth-Splatting, a method for estimating 3D states of cloth from RGB images through a prediction-update framework. Cloth-Splatting leverages an action-conditioned dynamics model for predicting future states and uses 3D Gaussian Splatting to update the predicted states. Our key insight is that coupling a 3D mesh-based representation with Gaussian Splatting allows us to define a differentiable map between the cloth state space and the image space. This enables the use of gradient-based optimization techniques to refine inaccurate state estimates using only RGB supervision. Our experiments demonstrate that Cloth-Splatting not only improves state estimation accuracy over current baselines but also reduces convergence time.
