Abstract
In this paper, we address the problem of tracking the tempo- ral evolution of arbitrary shapes observed in multi-camera setups. This is motivated by the ever growing number of applications that require consistent shape information along temporal sequences. The ap- proach we propose considers a temporal sequence of independently re- constructed surfaces and iteratively deforms a reference mesh to fit these observations. To effectively cope with outlying and missing geometry, we introduce a novel probabilistic mesh deformation framework. Using generic local rigidity priors and accounting for the uncertainty in the data acquisition process, this framework effectively handles missing data, rel- atively large reconstruction artefacts and multiple ob jects. Extensive ex- periments demonstrate the effectiveness and robustness of the method on various 4D datasets.