PT Unknown
AU Antonio Hernandez
   Miguel Reyes
   Sergio Escalera
   Petia Radeva
TI Spatio-Temporal GrabCut human segmentation for face and pose recovery
BT IEEE International Workshop on Analysis and Modeling of Faces and Gestures
PY 2010
BP 33–40
DI 10.1109/CVPRW.2010.5543824
AB In this paper, we present a full-automatic Spatio-Temporal GrabCut human segmentation methodology. GrabCut initialization is performed by a HOG-based subject detection, face detection, and skin color model for seed initialization. Spatial information is included by means of Mean Shift clustering whereas temporal coherence is considered by the historical of Gaussian Mixture Models. Moreover, human segmentation is combined with Shape and Active Appearance Models to perform full face and pose recovery. Results over public data sets as well as proper human action base show a robust segmentation and recovery of both face and pose using the presented methodology.
ER