CMP-VLID database

English French
videos count 832 485
mean frames count 4714 6508
meadian frames count 4291 4461
mean width x height 1189 x 673 1174 x 669

You can find links to the videos used for training and testing in the following archive: CMP-VLID.

Original paper:

Visual Language Identification from Facial Landmarks

Radim Spetlik, Jan Cech, Vojtech Franc and Jiri Matas

Last update: 12:38, 04/01/2017