KIT – University of the State of Baden-Wuerttemberg and National Research Center of the Helmholtz Association …continued Details and more Results Acknowledgment: This work was supported by the Quaero Programme, funded by OSEO; and BMBF contract no. 01ISO9052E and 13N12063. Data Set Statistics #characters, #tracks, #unknowns, #speaking speaking face assignment details for BBT Results Contact {baeuml, tapaswi}@kit.edu Project page (tracks, ground truth, etc.) http://cvhci.anthropomatik.kit.edu/projects/mma Identification Details obtain track scores by averaging frame scores … … and assign the most likely class (highest score) Assignment to unknown model all unknowns in a joint class use training data from tracks w/o name in the transcripts assignment as above Confusion Matrix for BUFFY-1..6 Failure Analysis BBT-1..6: BUFFY-1..6: