my lab
Fully automatic person identification in a video. Absolutely no manual labeling was done, instead our algorithm automatically learns a person model from a video and associated screenplay, from which we extract the set of characters appearing in a scene (bottom labels). For more details, see our CVPR 2009 paper. Toolboxes Convex Learning from Partial Labels Toolbox new Graph matching toolbox Multiscale NCut Image segmentation Toolbox Normalized Cuts Segmentation Code Datasets Annotated Faces on TV Datase
Timothee Cour , Praveen Srinivasan, Jianbo Shi. Balanced Graph Matching . Advances in Neural Information Processing Systems ( NIPS ), 2006 [ pdf ][ abstract ][ bib ] [ supplement : Affinely Constrained Rayleigh Quotients] [ code ]
Timothee Cour , Florence Benezit, Jianbo Shi. Spectral Segmentation with Multiscale Graph Decomposition . IEEE International Conference on Computer Vision and Pattern Recognition ( CVPR ), 2005. [ pdf ][ abstract ][ bib ] [ code ]
my lab