4D Cardiff Conversation Database (4D CCDb): a 4D database of natural, dyadic conversations.

The 4D Cardiff Conversation Database (4D CCDb) is the first 4D (3D Video) audio-visual database containing natural conversations between pairs of people. This publicly available database contains 17 conversations which have been fully annotated for speaker and listener activity: conversational facial expressions, head motion, and verbal/non-verbal utterances. It can be accessed at http://www.cs.cf.ac.uk/CCDb. In this paper we describe the data collection and annotation process. We also provide results of a baseline classification experiment distinguishing frontchannel from backchannel smiles, using 3D Active Appearance Models for feature extraction, polynomial fitting for representing the data as 4D sequences, and Support Vector Machines for classification. We believe this expression-rich, audio-visual database of natural conversations will make a useful contribution to the computer vision, affective computing, and cognitive science communities by providing raw data, features, annotations, and baseline comparisons.